Recent researchers proposed a cadence curl-up test over the traditional sit-up test (Sparling, Millard-Stafford, & Snow, 1997) for evaluating muscular endurance because the former test possesses logical validity, whereas the latter test is a speed test. However, others argued that the cadence curl-up test is impractical because most participants had difficulty in following the pre-determined cadence (Zhu, 1998). In other muscular endurance tests, such as the push-up test, cadence is not important as long as the participants perform the movement continuously. With this in mind, the question that if varying the cadence in curl-up test affects the evaluation of abdominal fitness of participant is presented. The purpose of this study was to evaluate the influence of cadence in curl-up test for evaluating (or classifying) abdominal fitness. A total of 93 high school boys aged from 15 to 18 (15.9+/-0.97) participated in a Georgia Tech curl-up test (GTCU) with 5 different cadences in a randomized order. The selected cadences were 20 reps/min, 25 reps/min, 30 reps/min, 35 reps/min, and free cadence (i.e., no required rhythm as long as the movement was performed continuously). Each trial of the GTCU test was given at least 7 days apart. The performance scores of each GTCU (number of repetitions) were then classified into quintile categories (low, fair, average, good, excellent). The Gamma statistic for agreement between ordinal variables (Siegel & Castellan, 1988, pp. 291-298) was employed to evaluate the agreement of fitness classification between each pair-wise cadence of GTCU. Results indicated that agreements of classification were fair to high (G coefficients ranged from .44 to .81), with an average G coefficient of .65 (.51<95CI<.79), which reflects satisfactory agreement was reached between different cadences. Although more repetitions were achieved for slower rhythm, the overall differences in mean reps between cadences were small. When same analyses were repeated for using 3-category classification (low, average, high), the average G inflated to .70. The classification intervals were similar across different cadences. When performance scores were classified into pass/fail categories (60%tile as cutoff), the average Phi coefficient was .44 (p<.001). In summary, the present study revealed that regardless of the speed of cadence, the classifications of abdominal fitness levels in high school boys were fairly consistent. It is concluded that, for practical reasons, the cadence free curl-up test is recommended for evaluating abdominal fitness of high school boys.Keyword(s): assessment, exercise/fitness, measurement/evaluation