Crop production in dryland systems is mainly dependent on water availability from rainfall which is highly variable between years and locations. We employed the widely used boundary-line analysis, with an existing industry dataset from across the Australian dryland cropping regions, to investigate the relative sensitivity of grain yield in canola (Brassica napus L.), chickpea (Cicer arietinum L.), faba bean (Vicia faba L.), field pea (Pisum sativum L.), lentil (Lens culinaris L.), and narrow-leafed lupin (Lupinus angustifolius L.) to variation in rainfall totals. Chickpea had the lowest non-productive water use, was more responsive to water supply, and reached its maximum yield at a lower water supply than the other species. In contrast canola had the highest non-productive water use, was less responsive to water supply, and reached its maximum yield at a higher water supply than the other species. These results suggest that chickpea offers the most stable outcome, and canola the greatest variation, in response to the variability in rainfall totals between years and locations.