Flourensia cernua DC. (tarbush) is a deciduous shrub with potential as a high-protein forage source for livestock. Twenty-four Polypay x Rambouillet wethers housed in metabolism crates were used to evaluate tarbush as a N source for sheep fed a low quality grass diet. Treatments were 100% ground tobosa grass (Pleuraphis mutica Buckl.) or tobosa substituted with 10, 20, or 30% whole pre-bloom tarbush leaves (n = 5) or 26% ground alfalfa (n = 4, Medicago sativa L.) on a dry matter basis (dmb). Sheep were fed ad libitum for 11 days, after which feed was restricted to 1 % (dmb) of body weight for 11 days to reduce sorting and maintain uniform intake. Apparent dry matter digestibility was not improved (P = 0.2646) with tarbush or alfalfa. Fecal N was similar (P = 0.1626), but urinary N varied (P = 0.0008) among treatments. Apparent N digestibility differed (P = 0.0042) among treatments (43, 46, 50, 56, and 63% for sheep consuming 0, 10, 20, or 30% tarbush or alfalfa, respectively). All treatments resulted in similar (P = 0.1569) but negative N retentions (-2.4, -2.2, -2.8, -2.0, and -1.5 g day^-1 for sheep consuming 0, 10, 20, or 30% tarbush or alfalfa, respectively). Serum clinical profiles (day 22) confirmed all sheep were nutritionally stressed, but did not indicate toxicosis. Although neither tarbush nor alfalfa N compensated for the low quality basal diet, N from 30% tarbush was utilized with similar efficiency to alfalfa N. The major impediment for using tarbush as a N source appeared to be low palatability.