Context : Maize (Zea mays L.)-based intercropping – growing maize with at least one species in the same field for a significant period – is a common practice in China. Intercropping may allow ecological intensification of maize production by achieving higher yields and higher resource-use efficiency than sole maize. Such advantages strongly depend on interactions between species and nitrogen availability. Unfortunately, there is a shortage of experiments combining both species choice and nitrogen management. Research question : It remains unclear how growth characteristics of the companion species, and notably its temporal complementarity with maize, and nitrogen fertilizer affect the performance of maize-based intercropping. Therefore, this work aims to explore the importance of the recovery response in relation to the duration of co-growth and that of maize recovery in interaction with nitrogen availability on the yield advantage of maize-based intercropping. Methods : A two-year field experiment was conducted in Gansu Province (China) to quantitatively determine the effect of different companion species and nitrogen fertilizer rates on yields, relative yields and land equivalent ratios (LER). Experimental treatments included two nitrogen input rates (120 and 240 kg N ha−1), and six companion species in a substitutive design: wheat (Triticum aestivum L.), linseed (Linum usitatissimum L.), cabbage (Brassica oleracea L.), garlic (Allium sativum L.), pea (Pisum sativum L.) and soybean (Glycine max L.). Results : Yields of intercropped maize were higher at 240 kg N ha−1 than at 120 kg N ha−1. Averaged over nitrogen inputs, the yield of maize intercropped with pea, cabbage, garlic, wheat, linseed and soybean was 71%, 72%, 78%, 53%, 50% and 58% of that of sole maize. LERs were greater than one for all species combinations, except for soybean/maize. LERs and overyielding of maize were higher at 120 kg N ha−1 than at 240 kg N ha−1. Overyielding of maize was negatively correlated to species co-growth duration and positively correlated to temporal niche differentiation, with the highest overyielding of maize with garlic corresponding to the shortest co-growth period. Conclusions : Overyielding was significantly increased with the duration of the recovery period of maize (after harvest of the companion species), while it was negatively correlated with the duration of the co-growth period. Our results suggest that the competition-recovery principle contributed greatly to yield advantage in maize-based intercropping. Implications : Our results should be considered in designing maize-based intercropping by choosing species and dates of sowing and harvesting for maximizing yield benefits.