The current study extended previous research on curriculum-based measurement in mathematics (M-CBM) assessments. The purpose was to examine the generalizability and dependability of multiple-skill M-CBM computation assessments across various assessment durations (1, 2, 3, 4, 5, and 6 minutes). Results of generalizability and dependability studies (N = 104 students) suggest that relative interindividual decisions can rely on the results from 1-minute administrations for low-stakes decisions and the results of 4-minute administrations for high-stakes decisions. Moreover, absolute intraindividual decisions can rely on the results from 4-minute administrations for low-stakes decisions and 13-minute administrations for high-stakes decisions. The implications and limitations of these results are discussed.