The effects of different cereal precooking process (roasting and extrusion cooking) on iron availability and protein digestibility of four African weaning foods were investigated using in vitro methods. In two weaning foods based on pearl millet, cowpea and peanut, the cereal was extruded (A) or roasted (B). In two other weaning foods having a similar composition, a low proportion of milk powder was added and the cereal extruded (C) or roasted (D). The mean +/- SD iron values (mg/100 g) were: A, 5.56 +/- 0.18; B, 9.12 +/- 0.93; C, 5.89 +/- 0.23; D, 9.04 +/- 0.85. When the pearl millet was roasted, the iron content was higher than in the extruded weaning foods (P < 0.01). However, the percent of available iron of the roasted weaning foods was very low (B, 1.64 +/- 0.01; D, 0.91 +/- 0.02). The iron availability of the extruded weaning foods, A and C, was 3.5 times and 6.5 times higher than the corresponding roasted weaning foods, B and D, respectively. This represented 332.4 +/- 4.4 and 375.1 +/- 5.8 micrograms of available iron/100 g for A and C, respectively versus 149.5 +/- 0.9 and 82.2 +/- 1.8 micrograms of available iron/100 g for B and D, respectively. No significant differences in polyphenol contents were found according to the precooking process of the cereal. The extruded weaning foods showed a higher protein digestibility of approximately 10% than the roasted ones (P < 0.05). A positive correlation was found between in vitro iron availability and protein digestibility (r = 0.976, P < 0.02). Despite a high content of iron, the iron availability of roasted pearl millet weaning foods was quite low. Extrusion cooking of the cereal improved the protein digestibility and iron availability of pearl millet weaning foods; however, the amount of available iron remained insufficient to meet the iron requirements of infants whatever the cereal processing.