We constructed a regression model to estimate the retention of electronic boluses in goats. With this aim, 2,482 boluses were administered to goats from dairy (Murciano-Granadina, n=1,326; French Alpine, n=381) and meat (Blanca de Rasquera, n=532) breeds. A total of 19 bolus types made of materials (ceramic, plastic tubes filled with concrete or silicone, and ballasts) differing in their specific gravity (SG) were used, thereby obtaining a wide variation in bolus features: diameter (9 to 22mm), length (37 to 84mm), weight (5 to 111g), volume (2.6 to 26mL), and SG (1.0 to 5.5). Each bolus contained a half-duplex glass encapsulated transponder (32 × 3.8mm) and was administered using adapted balling guns. Murciano-Granadina and Blanca de Rasquera goats also wore 2 visual plastic ear tags: V1 (double flag, 5.1g) and V2 (flag-button, 4.2g). No data on ear tags in French Alpine goats was available. Bolus and ear tag retention [(retained/monitored) × 100] was recorded for at least 1 yr. Dynamic reading efficiency [(dynamic reading/static reading) × 100] was also evaluated from 1,496 bolus readings. No administration incidences or apparent negative behavior or performance effects were observed for any bolus type. Static reading efficiency of retained boluses was 100%, except for the prototypes with metal ballasts, which yielded a 93.3% reading efficiency. Retention of metal-ballasted boluses was confirmed using x-ray equipment. Excluding ballasted boluses, a 99.5% dynamic reading efficiency was obtained. Ear tag losses were 6.5% for V1 and 3.7% for V2, ranging from 3.2 to 7.8% depending on ear tag type and goat breed. Bolus retention varied (0 to 100%) according to their physical features. Obtained data allowed the fitting of a logistic model of bolus retention rate according to bolus volume and weight (R2 = 0.98); the SG was implicitly considered. Estimated weight and SG to produce medium- (15mL) and standard-sized (22mL) boluses for 99.95% retention rate in goats were 73.0 and 94.1g, and 4.87 and 4.28, respectively. Mini-boluses are not recommended, as none of the available radio translucent materials reach the required SG. In conclusion, increase of SG was fundamental to optimize bolus retention. Medium-sized boluses (10 to 15mL; SG 5.8 to 4.9) for early administration and efficient retention in adult goats could be produced, thereby addressing the problem of using proper boluses for goat identification.