Improvements in the accuracy of single estimations of biochemical constituents such as the serum free hydroxyproline (FSHP) concentration can only result from control of biological variation as well as analytical variation. In this context two experiments were carried out to determine whether a significant diurnal rhythm in FSHP concentration exists in man, and whether a worthwhile reduction in intra-individual variance can be achieved by eliminating dietary sources of hydroxyproline. A significant (p < 0.001) diurnal variation in FSHP concentration was detected in a group of normal subjects, with a range of about 20% of the mean. In a second group of eight normal subjects, there was a substantial reduction in the within-subject variation after the start of dietary control. For the most precise determination of FSHP concentration therefore, dietary hydroxyproline sources must be controlled, preferably eliminated, and blood should be sampled at a standardised time of day.