Four, lactating dairy crossbreds ($50%{ imes}50%$ Holstein Friesian${ imes}$Native Zebu cattle) were randomly assigned according to a $2{ imes}2$ factorial arrangement (two protein levels and two levels of mangosteen peel pellets (Mago-pel)) in a $4{ imes}4$ Latin square design to receive four dietary treatments. All cows received concentrate at a proportion of 1 kg concentrate per 2 kg of milk yield, and urea-treated 5% rice straw (UTRS) was given ad libitum. It was found that total dry matter intakes, nutrient digestibility, ruminal pH and $NH_3$-N concentrations were not affected (p>0.05) by treatments. Concentrations of ruminal pH and $NH_3$-N were not affected by dietary treatments although the concentration of BUN varied significantly (p<0.05) between protein levels (p<0.05). The populations of rumen bacteria and fungal zoospores did not differ among treatments (p>0.05); however, the population of protozoa was decreased (p<0.05) when cows received Mago-pel supplementation. The composition of the population of bacteria, identified by real-time PCR technique, including total bacteria, methanogens, Fibrobacter succinogenes and Ruminococcus albus was similar (p>0.05) among dietary treatments (p>0.05); however, copy numbers of Ruminococcus flavefaciens was increased when protein level increased (p<0.05). Microbial protein synthesis, in terms of both quantity and efficiency, was enriched by Mago-pel supplementation. Milk yield was greatest in cows fed UTRS based diets with concentrate containing protein at 16% CP with Mago-pel, but were lowest without Mago-pel (p<0.05). In addition, protein level and supplementation of Mago-pel did not affect (p>0.05) milk composition except solids-not-fat which was higher in cows fed the diet with 19% CP. Therefore, feeding a concentrate containing 16% CP together with 300 g/hd/d Mago-pel supplementation results in changes in rumen fermentation and microbial population and improvements in milk production in lactating dairy crossbreds fed on UTRS.