In this study, we estimate production costs and elasticities of factor substitution for Zimbabwean smallholders, using a dual (cost function) approach with detailed data on prices paid and received by each of 65 farms across six survey sites over two years. We find that 95% of observed farm choices are consistent with optimal input use, and that there is moderate substitutability between labor, biochemical inputs and capital. These results indicate that farmers can substitute between factors as relative prices change, particularly to increase labor use as the rural population grows. By stratifying our sample, we investigate the degree to which production costs differ among the socioeconomic groups, tsting for higher costs among female-headed households (who might be subject to gender discrimination), resource-poor farmers without their own draft animals (who might have less timely operations), and isolated farms far from paved roads (who might have less access to markets and information). We find significant support only for the paved-roads effect, indicating the importance of rural infrastracture in determining production costs.