<p>Promoting gardening among urban residents holds the potential to improve urban diets in low- and middle-income countries, but there is a lack of evidence of impact. This study tests the hypothesis that training urban residents in gardening increases their intake of fruit and vegetables. It uses panel data for 254 control and 425 treatment households from four city corporations in the Dhaka metropolitan area of Bangladesh. Urban residents, 85% of whom were women, were interviewed before the start of an urban gardening program and one year after training and inputs were provided to the treatment group. The study estimated the average treatment effects using a difference-in-difference estimator. Of the 38 outcomes tested, 20 are significant (p &lt; 0.05) with 19 indicating a beneficial effect and one indicating an adverse effect. Among the beneficial effects, there is an increase in the diversity of fruits and vegetables produced (+ 5 species, p &lt; 0.01), the frequency of harvesting (+ 0.64 times/week; p &lt; 0.01), and increased sharing of produce with neighbors (+ 8%, p &lt; 0.01). Regarding food and nutrition, there is an increase in women’s dietary diversity score (+ 0.37 on a 0–10 scale; p &lt; 0.01), women’s minimum dietary diversity (+ 4%; p &lt; 0.01), and in the number of portions of cooked vegetables eaten (+ 0.96 portions/day; p &lt; 0.01). The gardening intervention also contributes to a range of perceived social, personal, and psychological benefits. The adverse effect is an increase in ultra-processed food consumption (+ 19%, p = 0.04). Nevertheless, the results confirm that urban gardening interventions can improve the quality of urban diets alongside other benefits.</p>