Managing free-roaming dog populations is a critical global public health issue, particularly in urban settings. Accurate population estimates are essential for designing effective management and rabies control strategies. This study aimed to estimate the free-roaming dog population in Klang Valley, Peninsular Malaysia, using multiple analytical models based on the photographic sight-resight method. Data-including GPS location, ownership status, age, and environmental factors-were collected across five districts and 15 towns between March and October 2022. Secondary data on land area, human population density, and ethnicity were also analysed. Five population models were applied to estimate free-roaming dog populations, assessing density per street length and area. A total of 599 dogs were recorded, comprising 492 sightings and 107 resights. The overall detection probability was 0.38 (95 % CI: 0.35-0.41), with population estimates ranging from 818 (Bailey method) to 1407 (Schnabel method). The Lincoln-Petersen index and Schnabel method yielded population estimates of 937 and 1407, respectively, but with low precision (27.53 % and 41.51 %). In contrast, the Bailey correction, Chapman correction, and Detection Probability models provided more precise estimates, with percentage precision values of 4.89 %, 3.53 %, and 2.39 %, respectively. The Detection Probability model emerged as the most precise, accounting for unseen individuals and detection bias-a crucial factor for accurate population estimation in free-roaming dog studies. Dog density per street length ranged from 3.33 dogs/km (direct count) to 8.34 dogs/km (detection probability). Overall population estimates varied significantly, ranging from 23,120 to 33,340 depending on the estimation method. Heatmaps revealed strong correlations between dog density, ethnicity, and environmental factors. These findings underscore the importance of precise estimation models to inform effective dog population management and rabies control strategies.