Honeycomb selection in the F2 generation of maize (Zea mays L.) hybrid PR 3183, based on line performance per se in the absence of competition, led to recycled hybrids with improved potential yield per plant (mean yield per plant in the absence of competition). In the present study six S6×S6 recycled hybrids and two commercial single-cross hybrids (PR 3183 and B73×Mo17) were tested at three plant densities (25000, 41667 and 83333 plants/ha), in two locations (Technological Education Institute farm of Florina, Greece and University farm of Thessaloniki, Greece), for 2 years (1998, 1999). The study was undertaken to assess indirectly the potential yield per plant (p), the crop yield potential (Ymax), and the optimum plant density (Dopt) of the hybrids. Estimate of p and Ymax were obtained through linear regression analysis of yield per plant (Yp) on plant density (D), expressed by the equation Yp = p−qD, with Ymax being equal to (1/4)p2q−1. Optimum plant density was assessed through linear regression analysis of natural logarithm of yield per plant on plant density, expressed by the equation ln(Yp) = α−bD, with Dopt being equal to 1/b. The recycled hybrids had higher estimated potential yield per plant (p), than the two check hybrids, with p values being positively correlated with yield per plant of hybrids obtained experimentally in the absence of competition. Results indicated that the higher potential yield per plant decreases the optimum plant density, and renders the hybrids less density-dependent.