Epidemiological studies in humans and experimental work in rodents indicate that diets high in fat and sugar are associated with mild cognitive impairment. Most animal studies modelling dietary effects on cognition have used continuous access protocols. However, most people do not eat unhealthy diets continuously but rather alternate between periods of eating well and eating less well. The present set of experiments evaluated the effects of intermittent, ‘cycling’ access to a western-style ‘cafeteria’ (CAF) diet on short-term place and object recognition memory in adult male rats. CAF comprised of a range of palatable supermarket foods in addition to regular chow. Experiment 1 compared a control group fed chow with groups fed CAF for 3, 5 or 7 days per week. Groups fed CAF for 5 or 7 days/week exhibited place recognition memory impairments and elevated fat mass, with no group differences in object recognition. Experiment 2 found that 3-day CAF cycles impaired place recognition memory regardless of the interleaving duration on chow (2 or 4 days). Experiment 3 fed all rats 16 days of CAF diet in blocks of 2, 4, 8 or 16 days, each separated by 4 days of chow. All four CAF schedules reduced place recognition relative to baseline; however, performance was worse when CAF was given as a single 16-day block than in eight 2-day blocks. All groups exhibited a comparable increase in fat mass. Thus, the pattern of access to high-fat, high-sugar diets influences place recognition memory impairments even when controlling for total diet duration. Larger ratios of days on versus days off CAF produce greater impairments, and time on the diet (rather than the time in between) may be the more critical component of this ratio. Thus, interspersing periods of healthy eating may mitigate the adverse effects of poor diets on the brain.