Trees in two provenance-progeny experimental sites of Pinus brutia Ten. were sampled to study variation in wood density and its components among and within six populations and to estimate heritability of wood density and ring components. Wood increment cores (12 mm thick) were collected from 29-year-old trees at breast height. A total of 1023 wood strips were scanned with the X-ray densitometry technique. There were significant differences among populations and among families (within populations) in all the characters studied. Murtbeli (M) population from the middle altitude (486 m asl) had the highest values in both ring-areaweighted density and late wood proportion (500 +/- 2.7 kg/m(3) and 49 +/- 0.3 %, respectively), whereas Hacibekar (H) population from the highest altitude (1032 m asl) exhibited the lowest values (468 +/- 3.0 kg/m(3), 42 +/- 0.3 %). Relatively high heritability values (hi(2) between 0.45 and 0.74; hhs 2 between 0.68 and 0.80) for investigated characters (except latewood density) suggest that these characteristics are under moderate to strong genetic control, and thus, by selecting populations (and families within populations) having high wood density, genetic improvement can be achieved in P. brutia. Correlations between wood density and other growth traits have indicated that simultaneous selection of desired genotypes for most of the wood density components is possible for the species. Latewood percentage and earlywood density present an optimal target for selection to improve wood density.