Employing data from an intercropping field experiment which measured the yields achieved by a surviving crop when the other crop failed at 0, 6 or 10 weeks after sowing, the use of a simple computer model to study the long-term yield effects of compensation in intercropping is described. Assuming different probabilities of crop failure for one or both of the crops, the model provided a comparison of long-term intercrop and sole crop yields from intercropping systems of oat–mustard, mustard–bean and oat–bean. It showed that if failure was restricted to only one of the crop species, intercropping suffered less yield decline with increasing probability of failure than did sole cropping only if there was a sufficiently high yield of the surviving crop. However, in the more realistic situation where each crop was subjected to the possibility of failure, this decline in yield was less in intercropping than sole cropping for all three combinations because poor yields when one species was the surviving crop were offset by good yields when the other species was the survivor. Mustard–oat and mustard–bean gave no yield advantage in the absence of failure (Land Equivalent Ratios less than 1) but they outyielded sole cropping for failure probabilities above about 5% and 15%, respectively. All combinations gave large intercropping advantages at high probabilities of failure. Oat–bean always had a strong yield advantage, even with no failure.

The model is very simplistic but it illustrates the potential for modelling long-term effects from field data. Limitations of the model are discussed and it is emphasized that these can be overcome if information to make the underlying assumptions more realistic is available.