Budwood from seven different clones of rubber tree [Hevea brasiliensis (Willd. ex Adr. de Juss.) Müell. Arg.] was planted in replicated trials in four different test sites in the plateau region of the state of São Paulo, Brazil. The clones represented a range of imported germplasm, whereas the locations were selected to represent a range of rubber growing areas. Girths were measured for six years consecutively, before the initiation of tapping for latex. Total number of latex vessel rings (lvr) and bark thickness (bt) were measured at six years. The largest average for all characters was observed in Matão. Overall means for girth, total number of latex vessel rings and bark thickness at six years were 35.26 cm, 11.30 units and 4.83 mm, respectively. Both the test sites and clones showed statistically significant differences in girth, lvr and bt values. There were significant clone x site interactions. Girth at one year was not a reliable predictor of future field performance. Broad sense heritability for girth was 0.16, and for lvr and bt it was 0.28 and 0.40, respectively, at six years. Repeatability was quite high, from 0.52 to 0.75 for all characters in all years. When two clones out of seven were selected, expected genetic gain in girth was about 2.0%; for lvr and bt it was 7.0% and 14.6%, respectively.
