Inclusive proton and deuteron spectra from 3He-induced reactions on 12C, 28Si and 58Ni have been studied. Each spectrum contains a continuum part that may be separated into a broad bump, centred around beam velocity energies at forward angles, and an exponential tail. The properties of each of these components have been investigated as function of outgoing particle, detection angle and target mass. The centroid position of the bump and the bump width are found to be almost independent of target mass and detection angle. The total (energy- and angle-integrated) cross sections for the bump show a mass dependence almost proportional to A2/3 for both protons and deuterons. The total cross section for the exponential tail has a different mass dependence for protons and deuterons: proportional to A and A1/3, respectively. The results for the proton tail are qualitatively reproduced by calculations assuming absorption of the projectile. Charged particle-proton coincidences have been measured and analysed in order to identify the major contributors to the inclusive proton and deuteron spectra at ¿ = 10°. The following processes have been identified: (i) elastic and inelastic breakup in which the projectile breaks up into two constituents leaving the target in its ground state or an excited state, (ii) absorptive breakup in which one part of the projectile continues virtually undisturbed while the other part is captured, and (iii) a process resulting in the preequilibrium emission of deuterons or tritons followed by the statistical emission from the residual compound system. These three processes account for 75–100% of the inclusive cross section. A semi-empirical model is presented that allows the calculation of the relative coincidence yields from the total inclusive cross sections. The total projectile breakup cross section amounts to about 20% of the total reaction cross section independent of target. A quasi-free breakup model is proposed for the description of the breakup spectra. The model reproduces the data qualitatively very well. The absolute cross sections are not reproduced since the absorption of the spectator is not included. The normalization constant for proton spectator processes are consistently a factor 4 larger than for the corresponding deuteron processes. It is suggested that this is related to the difference in transmission of the peripheral region of the nucleus.