Four cultivars of Hordeum vulgare L. were grown in 1967 at three seeding rates (54, 108, and 161 kg/ha) in six row spacings (11-, 18-, 23-, and 31-cm rows, and 11- and 18-cm cross-planted rows). The study was expanded in 1968 to include an additional cultivar of Hordeum distichum L. Grain yield was unaffected by seeding rate in either year. Narrow row spacing resulted in increased grain yields in 1968, when mean yields were high, but not in 1967 when yields were below normal. In both years, decreased row width resulted in increased numbers of spikes per m2 of ground area, and decreased numbers of grains per spike. A significant cultivar × row spacing interaction for yield was observed in 1968, but not in 1967. Statistical analysis revealed this interaction to be closely related to cultivar yield levels; that is, high yielding cultivars displayed a greater response to narrowing row spacing than did their lower-yielding counterparts. No consistent relationship was observed between cultivar yield differences, or the cultivar × row spacing interaction for yield, and cultivar differences in morphological type (leaf disposition, plant height). Furthermore, cultivar differences in yield were not related to the relative size of the various yield components (1000-grain weight, grains per spike, and spikes per m2). Although a cultivar × row spacing interaction for yield was observed, results suggest that, under Ontario conditions, this interaction does not represent a serious drawback to present cultivar testing procedures whereby cultivars are evaluated for yield at a single row width spacing.