The objective of this study is to investigate the application of the computational modelling methodology developed in the work of Bruzzi (PhD thesis. National University of Ireland, Galway, 2000) and Bruzzi and McHugh (Int. J. Fatigue 24 (2002) 1071) to predict the fatigue crack growth behaviour of two particle reinforced metal matrix composites (MMCs): (1) a forged 2124 Al reinforced with 17% SiC particles and (2) a cast 359 At reinforced with 20% SiC particles.The methodology described by Bruzzi and McHugh assumes a defect tolerant approach, using the effective range of the J-integral. DeltaJ(eff), to characterise the elastic-plastic fatigue crack growth (EPFCG) behaviour of an initial short crack. Experimentally determined crack closure levels are considered for the MMCs and predicted crack growth rate curves for an initial short crack are presented for varying stress levels, with a loading ratio, R = 0.1. Predicted S-N curves are then compared with experimental results for both the Al 2124 and Al 359 MMCs for various loading ratios. Good agreement is achieved. The fatigue limit versus defect size (Kitagawa diagram) is plotted for the Al 2124 MMC and excellent agreement is achieved in comparing the predicted results with analytical and experimental results. Finally, the predicted fatigue lives of the Al 359 MMC, using the defects sizes of the test specimens as determined by fracture surface analysis are compared to the actual experimental data. This validates the choice of initial defect size for confident prediction of the fatigue crack growth behaviour of the MMCs examined in this work. (C) 2003 Elsevier Science Ltd. All rights reserved.