The Transtheoretical Model (TTM), which asserts that health behavior change progresses in stages, is often used to explore health risk behaviors and to target and evaluate health promotion interventions. A four-question staging algorithm is often used to measure an individual's health behavior stage of change (SOC), but its accuracy or appropriateness for tailoring interventions or evaluating outcomes has not been established. The current study utilized data from three studies on HIV sexual risk behavior to compare SOC to reports of sexual risk on more detailed risk assessments, measured concurrently. Within each data set, detailed behavioral risk assessments were compared with SOC, with specific emphasis on maintenance staging, to evaluate the correspondence between SOC and reported behavior. Those classified in the maintenance SOC for condom use should, by definition, report no sexual risk events over the matched time period. Across all three studies, 18% of those classified in the maintenance SOC for condom use reported one or more sexual risk behaviors during the matched time period. Because the SOC algorithm is frequently used in intervention design, targeting, and evaluation, the potential for mis-categorization in the most advanced stage of maintenance raises concerns. Results suggest that intervention inclusion or evaluation strategies that use the maintenance stage as a primary outcome should be further qualified by behavioral data.