Biometric data on 897 Tibetan children living in Kathmandu, Nepal were collected over the period 1992 to 2000 with regular visits every 2 years. Measurements included cycloplegic autorefraction, A-scan ultrasonography, and video phakometry. Children who had not been studied at least once at age 12 years or older were not included in these analyses. The other subjects were divided into two groups; myopic if the refractive error was ever <-0.50 D, and emmetropic/hyperopic if the refractive error was never <-0.50 D, the nonmyopic group. Using all children who had been examined with four or five observations over time, the change of vitreous chamber depth with age by group was determined using a mixed-model regression method. The increase in vitreous length was 0.070 mm/year for the emmetropic group and 0.165 mm/year for the myopic group, with the differences apparent before the onset of myopia. An independent group of 59 children in whom there were two vitreous chamber depth measures before the age of 12 years and one measure taken after 12 years of age were used to assess the rate of increase in vitreous chamber depth as a predictor of myopia. Two other methods were examined using the independent group; the ratio of axial length to corneal radius of curvature and refractive error at age 10 years. Predictors based on rate of increase in vitreous chamber depth and axial length/corneal radius of curvature had sensitivities of 75% and 45%, respectively, and refractive error at age 10 years as a predictor for those who will not become myopic had a sensitivity of 88%.