Attempts to develop high yielding varieties in Ghana have mostly relied on the introduction of new clones to broaden the range of planting materials for yield improvement. The objective of this study was to estimate the genetic variation and heritability for bean yield of six recommended cocoa clones using these as males in crosses with five seed garden parents. Twenty-four families obtained from a 5 × 6 North Carolina II (NC II) incomplete factorial mating design together with 19 high yielding single crosses, of which four were standard mixed hybrids, were planted in a randomized complete block design with four replications and evaluated for bean yield over 6 years. To account for the serial correlation among yield data collected from the same plants over years, six models with different covariance structures were tested. The general covariance model emerged appropriate based on Akaike information criterion values with significant (P < 0.01) family × year interaction. Average bean yield was highest in the NC II families followed by the specific crosses then the standard mixed hybrids. Combining ability analysis among the NC II was significant for female, male and female × male interaction along with a narrow-sense heritability of 0.26. Clone CRG 6035 among the males which had good general combining ability could be added to the seed garden parents, while the promising hybrids (NA 33 × SCA 9, SCA 6 × Pound 10, T60/887 × PA 121 and T79/501 × CRG 6035/103) undergo stability tests before release.