Broiler production in semi-arid tropics must contend with high levels of heat stress, which have implications on stocking density, bird welfare, and profitability. Under these conditions, optimal stocking densities are likely to be lower than expected, thus must be experimentally determined. Therefore, this study investigated growth performance, haematology, serum biochemistry, carcass and meat quality, sizes of internal organs, and stress biomarkers in response to different stocking densities in broilers reared under semi-arid subtropical conditions. Five hundred, day-old Ross 308 broilers (44.0 ± 5.24 g live-weights) were randomly distributed to 25 replicate pens (1.32 m2 floor space each) to create five stocking densities: 1) 10 birds/pen (SD10); 2) 15 birds/pen (SD15); 3) 20 birds/pen (SD20); 4) 25 birds/pen (SD25); and 5) 30 birds/pen (SD30). There was a linear decrease (P < 0.05) in overall feed intake and weight gain in weeks 2 and 3 as stocking density increased. However, weight gain showed positive and negative quadratic responses (P < 0.05) in weeks 5 and 6, respectively, as stocking density increased. No linear or quadratic effects (P ˃ 0.05) were observed for overall feed conversion ratio, haematological parameters, and meat quality traits in response to stocking density. Symmetric dimethylarginine, alanine transaminase, and albumin levels quadratically increased (P < 0.05) in response to increasing stocking densities. Serum glucose and thigh weight were not affected (P < 0.05) while final body, drumstick, breast, and wing weights linearly declined with stocking density. Increasing stocking density linearly reduced (P < 0.05) the weights of gizzard, proventriculus, caecum, and colon. Stocking density had no effect (P ˃ 0.05) on latency-to-lie. It was concluded that higher stocking densities compromised feed intake, resulting in poor weight gains. Based on weight gain trends observed in week 5, it was determined that Ross 308 broilers should be reared at no more than 20 birds/pen (~15 birds/m2 or 27.27 kg/m2) under the experimental ambient conditions compared to the much higher globally accepted industry standard of 20 birds/m2.