This study evaluated the effectiveness of an automated observation and feedback system in improving safe sitting postures. Participants were four office workers. The dependent variables were the percentages of time participants spent in five safe body positions during experimental sessions. We used a multiple-baseline design counterbalanced across participants to test the system's effectiveness in delivering two types of feedback. For two participants, delayed/low-density feedback was introduced after baseline and immediate/high-density feedback was added in the next phase. For the other two participants, the sequence of implementing the feedback types was reversed. Results indicated that both delayed/low-density feedback and immediate/high-density feedback consistently improved participant postures and that immediate/high-density feedback was more effective than delayed/low-density feedback. The benefits of the automated system of postural data collection are discussed. (Contains 2 tables and 8 figures.)