This literature review addresses the technical adequacy of time sampling observation systems, which is fundamental to ensuring the generalizability of data obtained from behavioral observation. Investigations comparing partial interval, whole interval, and momentary time sampling have yielded consistent results. These suggest that duration of behavior is underestimated by whole interval sampling, overestimated by partial interval sampling, and on average accurately estimated by momentary time sampling. Momentary time sampling does not yield data suitable for the estimation of frequency. Partial interval sampling may yield consistent, but not necessarily absolutely accurate, estimates of frequency. Attempts to define the "ideal" interval for time sampling have generated inconsistent results. Investigations of momentary time sampling have yielded recommended values ranging from less than 30 seconds to 5 minutes. It appears that accuracy of momentary time sampling may be most dependent on overall frequency of observation, necessitating attention to both interval length and the length of the observational session. Little attention has been paid to issues of the representativeness of a behavioral sample drawn from a brief, fixed duration observation. The available literature suggests that massing observations within a fixed duration session may prove less generalizable than will observations spread over time. A list of 114 references is included. (Author/JDD)