OBJECTIVE To compare the results of noncycloplegic photorefraction, cycloplegic photorefraction and cycloplegic refraction in preschool and non-verbal children. METHODS One hundred and ninety-six eyes of 98 children (50 females, 48 males) were included in the study. Firstly, non-cycloplegic photorefraction was achieved with Plusoptix A09; secondly, cycloplegic photorefraction was carried out with Plusoptix A09 after 10 min cyclopentolate. Finally, 30min after instillation of twice cyclopentolate, cycloplegic refraction was obtained with autorefraction and/or standard retinoscopy. Spheric equivalent, spheric power, cylindric power and cylindrical axis measurements were statistically compared. RESULTS The mean age was 28.8±18.5mo (range 12-72mo). The differences in spherical equivalent, spheric power and cylindrical power measured by the three methods were found statistically significant (P<0.05). The spherical equivalent and spheric power measured by cycloplegic photorefraction were statistically higher than the measurements of the other methods (P<0.05). The cylindrical power measured by cycloplegic refraction was statistically lower than the measurements of the photorefraction methods (P<0.05). There was no significant difference in cylindrical axis measurements between three methods (P>0.05). CONCLUSIONS For the determination of refractive errors in children, the Plusoptix A09 measurements give incorrect results after instillation of cyclopentolate. Additionally, the cylindrical power measured by Plusoptix A09 with or without cycloplegia is higher. However, the non-cycloplegic Plusoptix A09 measures spheric equivalent and spheric power similar to cycloplegic refraction measurements in preschool and non-verbal children.