Annual forage yields of intercrops of berseem clover (Trifolium alexandrinum L.) and oats (Avena sativa L.), as affected by timing of initial harvest and sowing rate of oats, were investigated. Berseem clover was intercropped with oats at 0, 30, 60, 90 and 240 plants m )2 of oats in 1999 and 2000 in Alberta, Canada. Cutting date treatments involved initial harvest at 10-d intervals between 35 and 88 d after planting (DAP), and one or two subsequent harvests of regrowth. Total intercrop dry-matter (DM) yield averaged 12AE9-13AE3 t ha )1 with proportions of 0AE21-0AE43 of berseem clover in the forage. Treatments with an initial cut at the silage stage of oats (76 DAP in 1999, 88 DAP in 2000, at about soft-dough stage of oats) had greater yields of oats, lower yields of berseem clover, and lower proportions of berseem clover in total annual yields than treatments with an earlier initial cut. Total crude protein (CP) yield was greater with an initial cut at 65-66 DAP than with a silage-stage initial cut. With an initial cut at 35 DAP (before stem elongation of oats) or after 65 DAP (after heading of oats), yield potential of oats was sometimes reduced compared with silage-stage treatments, but this was balanced by a greater yield of berseem clover. The impact of harvest timing on total yield decreased as the proportion of berseem clover in the intercrops increased. With decreasing density of oats, DM yield of first-cut intercrops and total DM yield of oats decreased, while regrowth and total DM yields of berseem clover increased. Intercrops with oats at 60 plants m )2 had equal or greater total DM and CP yields than intercrops with 240 plants m )2 of oats. For intercrops with oats at 60 plants m )2 , with initial cuts at 65-66, 75-76, or 88 DAP, yields of regrowths were 0AE30-0AE35, 0AE16-0AE26 and 0AE09 of the total yield respectively. Oats-berseem clover intercrops showed potential to manipulate the pattern of annual forage yield and to provide flexibility of harvest without reducing annual yields.