A 7-year tillage experiment was conducted on a deep sand in the central wheat belt of Western Australia between 1998 and 2004 to evaluate the impact of tillage intensity [no-tillage (NT), conservation tillage (CT), and rotary tillage (RT)] on soil organic matter, microbial biomass and function, and crop yields in a wheat–lupin rotation. A fourth treatment (subterranean clover pasture, Pasture) with least soil disturbance was included as a comparison. By March 2004, total soil carbon (C) in NT and CT increased by 4.4 and 2.6 t/ha, respectively, to an average of 17.6 t/ha in the top 0.1 m of the soil profile. There was a loss of total soil C in RT (–0.5 t/ha), which was significant compared with the other 2 tillage treatments. Total soil C and nitrogen (N) contents in the pasture treatment were similar to those in NT and CT at the end of the experiment. Labile fractions of soil C responded more rapidly to tillage practice, with significant reductions by 2001 in light fraction C and dissolved organic C in the RT treatment compared with the other 3 treatments. The effect of RT on biology and function was seen early in the experiment and, compared with Pasture, NT, and CT, intense tillage in RT significantly reduced microbial biomass and cellulase activity in the surface 0.05 m by the third year of the experiment. However, at a depth of 0.05–0.10 m there were no significant differences between treatments. Grain yields in NT, CT, and RT were unaffected by tillage except in 2003, when lupin yield under RT (1.6 t/ha) was significantly less than under NT (2.0 t/ha) and CT (1.9 t/ha). Minimal differences between NT and CT are a reflection of the minimum disturbance in the CT treatment, although there were significant differences between CT and NT in microbial indices such as microbial quotient and metabolic quotient, suggesting a future divergence of these treatments.