This investigation tested the hypotheses that (1) physical workstation dimensions are important determinants of operator posture, (2) specific workstation characteristics systematically affect worker posture, and (3) computer operators assume "neutral" upper limb postures while keying. Operator head, neck, and upper extremity posture and selected workstation dimensions and characteristics were measured among 379 computer users. Operator postures were measured with manual goniometers, workstation characteristics were evaluated by observation, and workstation dimensions by direct measurement. Considerably greater variability in all postures was observed than was expected from application of basic geometric principles to measured workstation dimensions. Few strong correlations were observed between worker posture and workstation physical dimensions; findings suggest that preference is given to keyboard placement with respect to the eyes (r = 0.60 for association between keyboard height and seated elbow height) compared with monitor placement with respect to the eyes (r = 0.18 for association between monitor height and seated eye height). Wrist extension was weakly correlated with keyboard height (r = -0.24) and virtually not at all with keyboard thickness (r = 0.07). Use of a wrist rest was associated with decreased wrist flexion (21.9 versus 25.1 degrees, p < 0.01). Participants who had easily adjustable chairs had essentially the same neck and upper limb postures as did those with nonadjustable chairs. Sixty-one percent of computer operators were observed in nonneutral shoulder postures and 41% in nonneutral wrist postures. Findings suggest that (1) workstation dimensions are not strong determinants of at least several neck and upper extremity postures among computer operators, (2) only some workstation characteristics affect posture, and (3) contrary to common recommendations, a large proportion of computer users do not work in so-called neutral postures.