Abstract Selection of sorghum genotypes for the sorghum-cowpea intercrop system would be simplified if it could be done in sole crop. In order to compare evaluation in sole crop and in the presence of the standard cowpea cultivar C 152, sorghum inbred lines, F 1 hybrids and land races which differed in maturity date, height and canopy characters were grown in the two systems in two seasons at Hyderabad, India. Cowpea sole crop was included as an additional treatment. Sorghum canopy characters and yield components in intercrop were highly correlated with the same characters in sole crop. However, multiple regression of sorghum grain yield in intercrop on characters measured in intercrop explained more variation than regression on characters measured in sole crop. Characters related to light interception were the most influential in determining sorghum yield, but some genetically determined variation in yield was unexplained by either multiple regression. Characters related to light interception had a negative influence on cowpea yield, though again some variation due to sorghum genotype was unexplained. Thus although the influence of sorghum plant characters on each component crop is predictable, compensation between the components makes the overall outcome more difficult to predict, and dependent upon which component is favoured by the environment. The sorghum genotypes selected will therefore represent a compromise: they should not be dwarf types, but should be early maturing to escape drought, and have narrow canopies so as not to be too competitive on the cowpea. The final selection should be made in intercrop.