Rodeo Collection, Ltd. owns and operates the Rodeo Collection, an exclusive shopping center located on Rodeo Drive in Beverly Hills. Rodeo holds several registered service marks for the mark “Rodeo Collection” as used in connection with shopping center services. . In 1985, West Seventh and Statler and Waldorf announced plans to open a new shopping center in downtown Los Angeles under the name “The Collection.” Rodeo asked West Seventh to give the new shopping center a different name. West Seventh refused.

On February 5, 1986, Rodeo filed suit in district court alleging federal service mark infringement, federal unfair competition, and several related state claims. At the same time, Rodeo moved for a preliminary injunction to enjoin West Seventh from using any colorable imitation of Rodeo’s registered service mark to identify the new shopping center. The district court denied the motion. Rodeo timely appealed. We affirm.

DISCUSSION

1. Jurisdiction

Appellate jurisdiction is based on 15 U.S.C. § 1121 . The district court’s order is a reviewable interlocutory order. See Miss Universe, Inc. v. Flesher, 605 F.2d 1130 , 1131 n. 1 (9th Cir.1979).

2. Standard of Review

We will set aside the district court’s order denying Rodeo’s request for a prelimi *1217 nary injunction only if we conclude that the order was an abuse of discretion, was based on an erroneous legal standard, or was based on clearly erroneous findings of fact. See Bank of America National Trust and Savings Association v. Summerland County Water District, 767 F.2d 544, 547-48 (9th Cir.1985); Apple Computer, Inc. v. Formula International Inc., 725 F.2d 521, 523 (9th Cir.1984).

3. Standard Governing the Issuance of a Preliminary Injunction

To qualify for a preliminary injunction, the moving party must show either (1) a combination of probable success on the merits and the possibility of irreparable harm, or (2) that serious questions are raised and the balance of hardships tips sharply in the moving party’s favor. Sardi’s Restaurant Corp. v. Sardie, 755 F.2d 719, 723 (9th Cir.1985); Apple Computer, 725 F.2d at 523 . These are not two distinct tests, but rather the opposite ends of a single “continuum in which the required showing of harm varies inversely with the required showing of meritoriousness.” San Diego Committee Against Registration And The Draft v. Governing Board of Grossmont Union High School Disk, 790 F.2d 1471 , 1473 n. 3 (9th Cir.1986).

4. Probability of Success on the Merits

Rodeo alleges that West Seventh has infringed, and continues to infringe, the registered service mark “Rodeo Collection” by calling the new shopping center “The Collection.” Under 15 U.S.C. § 1114 (1), any person who uses in commerce a colorable imitation of a registered service mark, without the registrant’s consent and in a way that is likely to cause confusion, is liable for service mark infringement. The parties agree that “Rodeo Collection” is a registered service mark, that Rodeo is using it in commerce, and that West Seventh has adopted the mark “The Collection” for its new shopping center without Rodeo’s consent. Whether West Seventh’s mark is a colorable imitation of Rodeo’s mark is a pivotal factor in assessing the likelihood of confusion. Likelihood of confusion is thus the crucial issue. See Lindy Pen Co. v. Bic Pen Corp., 725 F.2d 1240, 1243 (9th Cir.1984). Rodeo’s federal unfair competition claim and its related state claims also turn on the likelihood of confusion. See Levi Strauss & Co. v. Blue Bell, Inc., 778 F.2d 1352, 1362 (9th Cir.1985) (en banc); Sardi’s Restaurant, 755 F.2d at 723 .

A likelihood of confusion exists not only when a consumer viewing a service mark is likely to purchase the identified services under the mistaken belief that they are the services of another service-provider, but also when the consumer would be likely to assume that the identified services are in some-way associated with another service-provider. See Alpha Industries, Inc. v. Alpha Steel Tube & Shapes, Inc., 616 F.2d 440, 443 (9th Cir. 1980).

We have developed a five-factor test for determining whether a likelihood of confusion exists:

See Park ‘N Fly, Inc. v. Dollar Park and Fly, Inc., 782 F.2d 1508, 1509 (9th Cir. 1986); Golden Door, Inc. v. Odisho, 646 F.2d 347, 349-50 (9th Cir.1980); Alpha Industries, 616 F.2d at 444 .

Likelihood of confusion requires that confusion be probable, not simply a possibility. HMH Publishing Co. v. Brincat, 504 F.2d 713, 717 (9th Cir.1974); Carter-Wallace, Inc. v. Procter & Gamble Co., 434 F.2d 794, 804 (9th Cir.1970). It is the totality of facts in a given case that is dispositive. See J.B. Williams Co. v. Le Conte Cosmetics, Inc., 523 F.2d 187, 190 (9th Cir.1975), cert. denied, 424 U.S. 913 , 96 S.Ct. 1110 , 47 L.Ed.2d 317 (1976); Fleischmann Distilling Corp. v. Maier Brewing Co., 314 F.2d 149, 160 (9th Cir.), cert. denied, 374 U.S. 830 , 83 S.Ct. 1870 , 10 L.Ed.2d 1053 (1963).

*1218 a. Strength of Rodeo ’s Service Mark

A mark’s strength can be measured in terms of its location along a continuum stretching from arbitrary, inherently strong marks, to suggestive marks, to descriptive marks, to generic, inherently weak marks. See Surgicenters of America, Inc. v. Medical Dental Surgeries, Co., 601 F.2d 1011, 1014-15 (9th Cir.1979).

A strong mark has secondary meaning. See Bada Company v. Montgomery Ward & Co., 426 F.2d 8 , 11 (9th Cir.), cert. denied, 400 U.S. 916 , 91 S.Ct. 174 , 27 L.Ed.2d 155 (1970). A secondary meaning arises if one of the connotations that a mark evokes is the particular good or service that it identifies. See 1 J.T. McCarthy, Trademarks and Unfair Competition, § 15:2 at 661 (2d ed. 1984). When first coined, descriptive marks have no secondary meaning, but over time, as customers begin associating them with particular goods or services, they may acquire secondary meaning. See Carter-Wallace, 434 F.2d at 800 .

Whether a mark is weak and descriptive or strong and distinctive can be determined only by reference to the goods or services that it identifies. Thus, “the mark BRILLIANT may be ‘descriptive’ on diamonds, ‘suggestive’ on furniture polish, and ‘arbitrary’ on canned applesauce.” 1 J.T. McCarthy, Trademarks and Unfair Competition, § 11:20 at 489 (2d ed. 1984).

Composite marks should be examined in their entirety, not piece by piece. See California Cooler, Inc. v. Loretto Winery, Ltd., 774 F.2d 1451, 1455 (9th Cir. 1985) (the validity of a composite mark is not to be judged by an examination of its parts); Franklin Mint Corp. v. Master Mfg. Co., 667 F.2d 1005, 1007 (CCPA 1981) (a mark should not be dissected and considered piecemeal).

We therefore analyze Rodeo’s mark in its entirety and in the context of shopping center services, not piecemeal and in the abstract. West Seventh’s argument that Rodeo’s mark is weak is based on a mistaken approach. West Seventh focuses on part of Rodeo’s mark, “collection,” and then argues that the overall mark is weak because the Los Angeles phone book reveals eleven other business uses of the term “collection.” Frequency of use, however, is not the proper standard. Instead, we must consider the specific associations that the composite mark triggers in the minds of consumers when it is used to identify shopping center services.

Two tests are commonly used to measure the strength of a mark, the “imagination test” and the “need test.” The “imagination test” focuses on the amount of imagination required in order for a consumer to associate a given mark with the goods or services it identifies. If a consumer must use more than a small amount of imagination to make the association, the mark is suggestive and not descriptive. See Golden Door, 646 F.2d at 350 ; AMF Inc. v. Sleekcraft Boats, 599 F.2d 341, 349 (9th Cir.1979). The “need test” focuses on the extent to which a mark is actually needed by competitors to identify their goods or services:

1 J.T. McCarthy, Trademarks and Unfair Competition, § 11:21 at 493 (2d ed. 1984). The two tests are related, because “[t]he more imagination that is required to associate a mark with a product [or service,] the less likely the words used will be needed by competitors to describe their products [or services].” Union Carbide Corp. v. Ever-Ready Inc., 531 F.2d 366, 379 (7th Cir.), cert. denied, 429 U.S. 830 , 97 S.Ct. 91 , 50 L.Ed.2d 94 (1976).

Both tests indicate that “Rodeo Collection” and the component term “Collection” are more than merely descriptive as used *1219 to identify shopping center services. A person would not be likely to picture a shopping center upon first hearing the name “Rodeo Collection.” One must use some imagination to associate the word “collection” with the collection of shops and restaurants that make up the shopping center. The remoteness of that association indicates that a competing shopping center would not need to use the term “collection” in order to identify its own shopping center.

b. Relationship Between the Identified Services

The fact that the two shopping centers are several miles apart and serve different clienteles is a factor in determining likelihood of confusion, but it is not controlling. To treat it as controlling would be to confuse the relationship between the consumers of the respective shopping centers’ services with the relationship between the services themselves. It is the latter relationship that we must examine. See Park ‘N Fly, 782 F.2d at 1509 ; Golden Door, 646 F.2d at 350 .

The competing services need not be identical to entitle the holder of the registered mark to protection; the zone of service mark protection may extend into the area of noncompeting services if the junior user’s mark may create the impression that the services identified are affiliated with the holder of the registered mark. See Fleischmann Distilling, 314 F.2d at 155 n. 6; King Research, Inc. v. Shulton, Inc., 324 F.Supp. 631, 637 (S.D. N.Y.1971). In this case, the two marks identify virtually identical services related to the leasing, management, and general operation of shopping center space.

c. Similarity Between the Marks

The district court made no finding regarding the similarity between Rodeo’s registered mark and West Seventh’s unregistered mark. We assess the similarity of the marks in terms of their sight, sound, and meaning. AMF, 599 F.2d at 351 . In these terms, the two marks are substantially similar, sharing as they do the substantive term “collection.” Their similarity is especially close in light of the unusualness of the word “collection” as an identifier of shopping center services. The two marks’ differences in lettering and design are insignificant; Rodeo’s registration of its mark in a plain format gives Rodeo rights to the mark in all forms of lettering. See Piccolo Sportswear, Inc. v. Mast Industries, Inc., 227 U.S.P.Q. 710 , 711 (TTAB 1985). Weighing the marks’ similarities more heavily than their differences, see AMF, 599 F.2d at 351 , we conclude that they are substantially similar. See generally Specialty Brands, Inc. v. Coffee Bean Distributors, Inc., 748 F.2d 669, 673 (Fed. Cir.1984) (“Spice Islands” as used to identify tea is confusingly similar to “Spice Valley” as used to identify tea); The Villager, Inc. v. Dial Shoe Co., Inc., 256 F.Supp. 694, 700 (E.D.Pa.1966) (“The Villager” as used to identify shoes is confusingly similar to “Miss Villager” as used to identify shoes).

d. Evidence of Actual Confusion

The district court found that Rodeo presented no evidence of actual confusion. Rodeo does not dispute this finding. Instead it argues that the lack of evidence of actual confusion is not controlling. We agree. Evidence of actual confusion is strong evidence of likelihood of confusion, but it is not determinative. See Park ‘N Fly, 782 F.2d at 1509 ; New West Corp. v. NYM Co. of California, Inc., 595 F.2d 1194, 1201 (9th Cir.1979).

e. West Seventh’s Intent in Adopting its Mark

The district court made no express finding concerning West Seventh’s intent. Evidence of wrongful intent, when present, has some bearing on the likelihood of confusion. The absence of such evidence, however, is not determinative. See Park W Fly, 782 F.2d at 1509 ; New West, 595 F.2d at 1201 .

*1220 f. Was There a Likelihood of Confusion?

We review the district court’s ruling that Rodeo failed to demonstrate a likelihood of confusion for clear error. Levi Strauss, 778 F.2d at 1355-56 . Rodeo has presented arguments that tend to indicate (1) that the service mark “Rodeo Collection” possesses secondary meaning as an identifier of shopping center services, (2) that the competing service marks identify the same type of service in the same general geographical area, and (3) that the competing service marks are substantially similar in sight, sound, and meaning. Rodeo made no showing regarding the other two elements of the five-part test: actual confusion and the alleged infringer’s wrongful intent. We recognize that such a showing is not required, but in this case the absence of such evidence, coupled with the showing made by West Seventh in response to Rodeo’s motion, precludes us from concluding that the district court clearly erred in holding that Rodeo failed to establish a likelihood of confusion.

5. Probability of Irreparable Harm

Once the plaintiff in an infringement action has established a likelihood of confusion, it is ordinarily presumed that the plaintiff will suffer irreparable harm if injunctive relief does not issue. See Apple Computer, 725 F.2d at 526 . The record in this case, however, precludes application of this presumption. Rodeo has made no independent showing of irreparable harm. See Sardi’s Restaurant, 755 F.2d at 725 .

The district court’s order denying Rodeo’s motion for a preliminary injunction was not based on an erroneous legal standard or on clearly erroneous findings of fact, nor was it otherwise an abuse of discretion. See Bank of America National Trust and Savings, 767 F.2d at 547-48 . In so holding we draw no conclusion regarding Rodeo’s ability to prove its case at trial.

AFFIRMED.