We have analyzed, by duplicate neutron-activation analysis, 125 members of iron meteorite groups IAB and IIICD. These data show no hiatus between the groups, and we recommend that the two sets be treated as a single group until data are obtained that require their separation. In cases where there is no ambiguity, we will use IAB to designate the combined group. Our data allow properties of group IAB to be more tightly constrained than heretofore. We examined the properties of ungrouped irons occupying the same region of GaNi and GeNi space as IAB and found that our more precise data did not add any to the group. Based primarily on CoNi, AuNi, and to a lesser extent, CuNi and AsNi trends we find that three irons (Hassi-Jekna, Magnesia and Qarat al Hanash) previously assigned to IIICD are better designated ungrouped, primarily because their Au contents diverge from those of IAB; six irons (EET 84300, Mertzon, Misteca, Persimmon Creek, Yongning, and Zacatecas (1792)) have also been removed from IAB on the basis of the new data. Fractional crystallization models of magmas saturated in troilite yield trends for several elements quite similar to those observed in IAB and, with moderate modification of the distribution coefficients, these can also account for the IIICD trend. However, this model requires metal segregation and convective stirring at very low temperatures, and predicts a much lower abundance of low-Ni irons than observed. It also fails to account for the ubiquitous presence of trapped melt and chondritic (or subchondritic) silicates in IAB irons. The weight of the evidence supports an impact-melt model in which individual IAB irons are interpreted as melt pools produced by impacts into a chondritic megaregolith. The lower the melt temperature of the melt, the larger the fraction of Ga, Ge, and Ir that remained sequestered in unmelted solids. The increasing range in Ga, Ge, and Ir with increasing Ni content can be explained by mixing between different primary melt pools having compositions along the lower (IIICD) envelope of IAB-IIICD. And we suggest that some high-Ni melt pools having low cotectic temperatures experienced crystal settling on a scale of cm to m; specifically, this process may be responsible for producing the most Ni-rich members of the group. We infer that the IAB precursor materials had properties (high porosity, fine grain size) that made them susceptible to impact melting. A high FeS would also favor melt generation. The oxygen isotope composition of IAB silicates is 0.45%o below the terrestrial-fractionation line in a region occupied by CR chondrites and other carbonaceous-chondrite-related meteorites. This suggests that the chondritic precursor was a metal- and FeS-rich carbonaceous chondrite. Cl-normalized element/Ni ratios in low-Ni (65–68 mg/g) TAB are ≥1.5 for Co, Ga, Ge, and Au. Although the high Ga and Ge abundance ratio could indicate fractional crystallization of elements having high solid/melt distribution coefficients, this explanation cannot account for the Co and Au enrichments. We interpret the high Ga, Ge, and Au ratios in terms of a nebular condensation-accretion model; we suggest that these elements condensed into a late-formed, fine nebular component having an enhanced abundance in the chondritic materials parental to group IAB. Because it is more refractory than Fe and Co, the Ni content of this fine metal was appreciably lower than that of the early formed coarse metal.