Raw whole milk of high microbial quality ([les ]4×104 cfu/ml) was processed using a ~15 l/h homogeniser with a high pressure (HP) valve immediately followed by cooling heat exchangers. The effects of homogenisation between 100 and 300 MPa (HP valve) with an initial milk temperature Tin=4 °C or 24 °C was investigated on the inactivation of: (i) endogenous alkaline phosphatase (ALP); (ii) endogenous milk flora and (iii) two Gram positive (Listeria innocua and Micrococcus luteus) and one Gram negative (Pseudomonas fluorescens) strains inoculated into milk. Temperatures T1 and T2 measured before and immediately after the HP valve, and fat globule size distributions were also determined. ALP activity slightly decreased after homogenisation above 250 MPa when Tin=4 °C (corresponding T2>58 °C), but markedly decreased above 200 MPa when Tin=24 °C (T2>60 °C). In contrast to inactivation induced by continuous short-time thermal treatments, ALP inactivation induced by HP homogenisation was clearly due to mechanical forces (shear, cavitation and/or impact) in the HP valve and not to the short ([Lt ]1 s) residence time at temperature T2 in the same valve. Inactivation of the three exogenous microorganisms led to similar conclusions. Homogenisation at 250 MPa or 300 MPa (Tin=24 °C) induced a 2–3 log cycle reduction of the total endogenous milk flora and a 1·5–1·8 log cycle reduction of inoculated List. innocua. Higher reduction ratios (2–4 log cycles) were obtained for the two other microorganisms. The highest levels of ALP inactivation corresponded to the highest extents of microbial reduction. Running the milk twice or three times through the homogeniser (recycling), keeping temperature T1≈29 °C and pressure=200 MPa, increased homogenisation efficiency.