BACKGROUND: Despite a doubling of HIV testing coverage in Kenya over the past decade, approximately 2 in 10 people with HIV remained unaware of their infection in 2018. HIV testing is most effective in identifying people with undiagnosed HIV through frequent and strategic testing in populations at high risk. An assessment of testing frequency and predictors of first-time and repeat testing is critical for monitoring effectiveness of testing strategies. METHODS: We conducted a cross-sectional analysis of adults aged ≥18 years who tested HIV-positive at 4 HIV testing and counseling clinics in Kenya from February 2015 to February 2016. We categorized individuals based on testing history, used Wilcoxon rank-sum tests to assess differences in intervals between the most recent and current HIV test, and used log-binomial regression to determine characteristics associated with first-time and repeat testing. RESULTS: Among 1136 people testing HIV-positive, 336 (30%) had never tested before and 800 (70%) had, of whom 208 (26%) had previously tested positive. Among previously negative repeat testers, the median intertest interval was 414 days in key/priority populations (interquartile range = 179-1072) vs. 538 in the general population (interquartile range = 228-1299) (P = 0.09). Compared with previously negative repeat testers, being a first-time tester was associated with being age ≥40 years [vs. 18-24; adjusted risk ratio = 1.67, 95% confidence interval (CI): 1.23 to 2.26], men (vs. women; adjusted risk ratio = 1.45, 95% CI: 1.21 to 1.71), and testing through provider-initiated testing and counseling (vs. client initiated; 1.19, 95% CI: 1.00 to 1.40). CONCLUSIONS: There is a need to increase HIV testing among older individuals and men, increase testing frequency in key/priority populations, and maintain provider-initiated and facility-based testing to reach first-time testers.