The NCII design (North Carolina mating design II) has been widely applied in studies of combining ability and heterosis. The objective of our research was to estimate how different base populations, sample sizes, testcross numbers and heritability influence QTL analyses of combining ability and heterosis. A series of Monte Carlo simulation experiments with QTL mapping were then conducted for the base population performance, testcross population phenotypic values and the general combining ability (GCA), specific combining ability (SCA) and Hmp (midparental heterosis) datasets. The results indicated that:

1. increasing the number of testers did not necessarily enhance the QTL detection power for GCA, but it was significantly related to the QTL effect.

2. The QTLs identified in the base population may be different from those from GCA dataset. Similar phenomena can be seen from QTL detected in SCA and Hmp datasets.

3. The QTL detection power for GCA ranked in the order of DH(RIL) based > F₂ based > BC based NCII design, when the heritability was low.

The recombinant inbred lines (RILs) (or DHs) allows more recombination and offers higher mapping resolution than other populations. Further, their testcross progeny can be repeatedly generated and phenotyped. Thus, RIL based (or DH based) NCII design was highly recommend for combining ability QTL analysis. Our results expect to facilitate selecting elite parental lines with high combining ability and for geneticists to research the genetic basis of combining ability.