Post occupancy evaluation is part of a design-evaluation-design cycle in which designers learn from their successes and mistakes and subsequently improve their designs. But, if designers want to make most effective use of information collected in such studies they must be done reliably and validly—few studies give evidence to justify such a claim. In the present study, two commonly and interchangeably used POE observation methods (direct observation and time-lapse photography) were comparatively tested in order to assess their reliability. Reliability concerns the extent to which different observers or the camera yield the same results in observing the same situation. The test case was conducted in a heavily used urban space and much of the data, from observer to observer, and observer to camera, was found unreliable. Reliability decreased as pedestrian frequency increased but not so uniformly that data from this study could be used to determine an exact number of persons that can be accurately mapped. Reliability "checks" should be made in pretesting of direct observations, also in retrieval of data from film. Direct observation and time-lapse photography can be used conjointly with the intent of using camera as an accurate basis against which to assess the reliability of direct observations, but with precaution taken to ensure the accuracy of camera data. Standards of reliability and validity, with simple tests or approaches to measuring them need to be developed in order to make it easier for researchers to “check” the reliability and validity of their findings.