Sixteen multiparous Holstein-Friesian crossbred cows in mid-lactation were blocked according to days in milk (DIM) and previous lactation and randomly assigned according to a Randomized Complete Block (RCB) design with four replications to receive four dietary treatments. The dietary treatments consisted of T1: No cassava hay (CH) or stylo 184 hay (SH) supplementation, supplementation of concentrate to milk yield at 1:2 (control), T2: Supplementation of 1 kg of CH/hd/d, supplementation of concentrate to milk yield at 1:2, T3: Supplementation of 1 kg of CH+SH/hd/d, supplementation of concentrate to milk yield at 1:2, T4: Supplementation of 2 kg of CH+SH/hd/d, supplementation of concentrate to milk yield at 1:3. All animals received Ruzi grass from a cut-and-carry system as roughage source. The feeding trial lasted for 9 weeks. The results revealed that DMI of concentrate of supplemented treatments were significantly lower (p<0.05) than those in the control, but there was no significant difference between T2 and T3. There was no significant difference in forage DM intake between the control and supplemented treatments. CP and NDF digestibility of supplemented treatments were significantly (p<0.05) greater than the control and there were no significant differences among supplemented treatments. Milk yield and 3.5% FCM (14.3, 14.5, 14.7 and 14.8; 13.9, 14.3, 14.3 and 14.6 kg/hd/d, respectively) were not significantly different among treatments. Milk protein percentage of supplemented treatments was significantly (p<0.05) higher than the control, but there were no significant differences among supplemented treatments. There was no significant difference in milk fat percentage between the control and supplemented treatments. However, milk fat percentage tended to be higher for supplemented animals as compared to the control group. There were also no significant differences in lactose, solids-not-fat and total solids percentages among treatments. Cows in supplemented treatments gave incomes over supplement cost (IOSC) of 2.72, 2.74 and 2.93 US$/hd/d, respectively which were greater than for cows on control treatment. Furthermore, IOSC were greatest for cows in T4 as compared to other treatments. Based on this study it was concluded that, feeding cassava hay solely or in combination with stylo 184 hay as a supplemental protein source could be a potential valuable strategy in small-holder dairy farming systems in the tropics. This strategic supplementation significantly reduced concentrate use, which resulted in improved milk yields and milk quality for the supplemented cows. Moreover, it resulted in higher economical returns through increased productivity and lower ratios of concentrate to milk yield, from 1:2 to 1:3. (Asian-Aust. J. Anim. Sci. 2004. Vol 17, No. 5 : 670-677)