This study examined the interrater reliability and temporal stability of a scoring system developed by Troyer, Moscovich, and Winocur [Neuropsychologia 11 (1997) 138] to measure clustering and switching on verbal fluency (VF) tasks such as the Controlled Oral Word Association Test (COWAT) [Benton, A.L., Hamsher, K., & Sivan, A.B. (1983). Multilingual aphasia examination (3rd ed.). Iowa City, IA: AJA Associates]. Seven independent raters scored COWAT protocols of 125 healthy participants in accordance with the rules proposed by Troyer et al. Intraclass coefficients were near perfect, ranging from .96 for total number of clusters to .99 for total number of switches. Test–retest reliability coefficients (n=55) were poor to modest (r=.47 for clusters, r=.58 for switches, and r=.70 for total words). Significant improvement in performance was observed across most COWAT indices, suggesting a practice effect. Modifications to test administration are suggested to improve the stability of cluster and switch scores, as well as other variables for further study.