The objectives were to determine browse species preference of goats using dry matter intake (DMI) as a proxy, to compare preference when offered in combination with polyethylene glycol (PEG) and to establish relationships between browse species intake and chemical compositional data. Air-dried leaves of Acacia etbaica, Cadaba farinosa, Capparis tomentosa, Dichrostachys cinerea, Dodonaea angustifolia, Euclea racemosa, Maerua angolensis, Maytenus senegalensis, Rhus natalensis and Senna singueana were used. Two cafeteria trials, each lasting 10 days, were conducted using four local mature male goats of 2-2.5 years receiving a daily ration of grass hay (4% of body weight) and 200 g wheat bran. In trial 1, goats were offered 25 g of each browse species for a total of 30 min with intake, time spent on consumption and the number of visits to specific browse species recorded at 10-min intervals. In trial 2, the same procedure was followed except that 25 g of PEG 4000 was added to the daily wheat bran ration. Crude protein and neutral detergent fibre in browse species ranged from 69.0-245.5 to 159.8-560.6 g/kg dry matter (DM) respectively. Total phenols and total tannins contents ranged between 3.7-70.6 and 2.5-68.1 mg tannic acid equivalent/g DM, respectively, and condensed tannins 1.7-18.4 Abs /g DM. Preference indicators measured in the first 10 min of browse species intake differed significantly among browse species and with PEG (p < 0.0001). Principal components explained 69.9% of the total variation in browse species DMI. Despite the high tannin levels, D. cinerea, R. natalensis and A. etbaica were the most preferred species regardless of PEG presence. Tannin levels at the observed browse species DMI did not determine preference, instead, preference appeared to be based on hemicellulose. Determining browse species preference is essential to exploit them to improve nutrient utilization and control parasites in goats.