Six Egyptian cotton varieties were set up in diallel cross mating design to produce 15 F1 hybrids and 15 F1r reciprocal hybrids at 1996 growing season. In 1997 growing season, two experiments were carried out at both Sakha, Kafr El-Sheikh Governorate (L1) and Kafr Saad, Damietta Governorate (L2). Each experiment was designed in a randomized complete blocks with four replications. The 36 entries were evaluated for earliness and fiber traits : First fruiting node (F.F.N.), days to first opening boll (D.F.O.B.), earliness index (E.I.), fiber fineness (F.F.), fiber strength (F.S) and fiber length at (2.5% S.L) under two location. The results of the analyses of variance indicated that differences among population means were present and highly significant for all studied traits at the two locations and combined data except (D.F.O.B) at (L1). Also the genotypes by location interactions were significant or high significant for all studied traits except fiber fineness (F.F) trait, indicating the presence of interaction between genotypes and locations. The lowest amount of heterosis versus mid– parents was (-0.64 %) for days to first opening boll and the largest amount was (4.60 %) for earliness index. Heterosis versus better–parent was not of economical importance. Tests of significant of the mean squares of general combining ability (G.C.A.) were larger in magnitudes than those specific combining ability (S.C.A.) and showed highly significance for all studied traits. In the same time specific combining ability by location variance (s2 s L) were larger than the other genetic components general combining ability by location variance (s2 g L) and reciprocal by location variance (s2 r L) for all studied traits. The magnitudes of dominance genetic variance (s2D) were larger than those additive genetic variance (s2A) and reciprocal genetic variance (s2r) for all studied earliness and fiber traits with the except of earliness index (E.I) trait.The estimates of broad sense heritabilities were larger in magnitude than their corresponding estimates of narrow sense heritabilities for all studied traits at two locations and from combined data except (F.F.N.) and (D.F. O.B) traits at location (L1), (F.S.) trait at (L2) and (E.I.) trait from combined data.