In low-input rice-maize rotation systems in the hills of central Madagascar, farmers deal with erratic rainfall, poor soils, high soil erosion risks and infestation by the parasitic weed Striga asiatica (L.) Kuntze. Practices combining zero-tillage with permanent soil cover by intercropped legumes and crop residue mulches — known as Conservation Agriculture (CA)— are proposed as remedy against soil and climatic constraints. Implications of these practices for S. asiatica are unknown. A 4-season factorial experiment compared the current farmer practice of rice - maize rotation, involving seasonal tillage and crop residue removal (CONV), with three rice - maize rotation systems following CA with different cover crops, i.e. Vigna unguiculata (cowpea) and Mucuna pruriens (CACM), Vigna umbellata (ricebean) (CARB), and Stylosanthes guianensis (CAST). Performance of two rice varieties, NERICA-4 and -9, with partial S. asiatica resistance, were compared with the locally popular B22. Parasite emergence time, numbers, and seed bank sizes were recorded. In all CA practices S. asiatica infection was significantly reduced. Best results were obtained with Stylosanthes guianensis (CAST). This species also suppressed ordinary weeds much better than other cover crops. With CAST, average parasite emergence was delayed by 7.5¿days (in rice) and 6.3¿days (in maize) and infection levels were reduced by 79% (in rice) and 92% (in maize) compared to the conventional farmer practice (CONV). NERICA varieties delayed S. asiatica emergence by 5.7¿days (NERICA-9) and 9.7¿days (NERICA-4) and reduced infection levels by 57% (NERICA-9) and 91% (NERICA-4) compared to B22. In maize the residual effect of resistance of NERICA-4 resulted in a delay of 7.5¿days in S. asiatica emergence and a reduction of 60% in parasite numbers. The best combinations delay S. asiatica emergence by 17.8¿days (CAST¿+¿NERICA-9) and 19.1¿days (CARB¿+¿NERICA-4) and reduce the parasite infection levels by 96% (CAST¿+¿NERICA-