1. To study the relationship between accommodation under natural viewing conditions, age and refractive errors, we have measured time courses of accommodation in thirty-nine human subjects aged 5-49 years using a newly developed technique. The technique is based on infrared photoretinoscopy and involves fully automated on-line image processing of digitized video images of the eyes with a sampling rate of 5.3 Hz. 2. The distance between the subject and the video camera was about 1.3 m. Head movements of the subject required little restriction because the eyes were automatically tracked in the video image by the computer program. All subjects were tested under binocular viewing conditions. 3. Both refraction of the right eye and pupil diameter were measured with a precision of 0.2-0.4 dioptres (D) and 0.1 mm, respectively, and were plotted on-line. The data were subsequently automatically analysed. 4. Automated infrared photoretinoscopy proved to be very convenient and easy to handle in both children and adults. 5. The maximal speed of accommodation for a target at a distance of 5 D declined in the subjects with age (from up to 21.7 D s-1 for accommodation and 32.7 D s-1 for subsequent accommodation to a distant target ('near to far accommodation') in children down to 2-18 D s-1 in adults). There was a striking inter-individual variability in the maximum possible speed of accommodation and near to far accommodation. 6. Speed of accommodation and of near to far accommodation was correlated for each subject. However, in most of the subjects, the process of near to far accommodation was faster than accommodation (P < 0.005, if averaged over all subjects). This correlation was independent of age. 7. The accommodation-induced pupillary constriction (pupillary near response) was absent in children for a 4 D target; even at 10 D, there was no reliable pupillary response. The pupillary near response increased to about 1.6 mm D-1 of accommodation at the age of 47. Since a pupillary near response could still be elicited in presbyopic subjects unable to accommodate, the ratio of pupillary constriction per dioptre of accommodation approached infinity. 8. The magnitude of the pupillary near response was highly variable even among subjects of the same age but was typical for each subject. There was a correlation (P < 0.01) to refractive error: corrected myopes had weaker pupillary near responses than emmetropes or hyperopes.(ABSTRACT TRUNCATED AT 400 WORDS)