Observation statements are characterized as being: (1) dependent upon the human senses or other sensory apparatus; (2) specific descriptions' (3) made by speakers who offer as the primary support for their statements the events or things which stimulated their making the statements; (4) used as the foundations for knowledge; and (5) easily believed by experts in the field in question. Thirty criteria or principles for critically judging the reliability of such statements are presented. First, observation statements are more reliable than the inferences based upon them. Second, certain observer characteristics affect reliability: emotionality; alertness; conflict of interest; skill with thing observed; theoretical understanding; normal senses; reputation; precision; skill with observation technique; and bias. Third, characteristics of the observation condition affect reliability: the instrument precision; quality; condition; range of application; and ease of understanding. Finally, reliability of the observation statement depends upon characteristics of the statement itself: commitment of speaker; corroboration or documentation; precision; fit into body of knowledge; and basis upon record. The statement should be made and believed by the observer, close to the time of observation. These criteria provide theoretical support for a test or curriculum on judging the reliability of observation statements. (CP)