The aim of this study was to evaluate carcass yield of surubins Pseudoplatystoma spp. reared in ponds under different stocking densities. We performed a completely randomized design, with three treatments (density) and nine repetitions. The densities were: density 1 (1peixe/5m2), density 2 (1 fish/3m2) and 3 density (1 fish/2.14 m2). Surubins total average weight of 1.144 ± 0.340 kg, were slaughtered by thermal shock in ice for determining yields of eviscerated fish with and no head (YEFWH; YEFWNH), the lateral fillet with and with no skin (YLFWS; YLFWNS) of abdominal fillet with and with no skin (YAFWS; YAFWNS), head (YH), skin (YS), viscera (YV) of residues (YR). Data were subjected to analysis of variance and average was compared by Tukey test at 5%. In order to steeming the relation among income on percentage of carcass components depending on the final weight of the fish we reckoned regression equations. There was no difference (P>0.05) of treatment through the variables of income. Average values for YEFWH, YH, YLFWNS, YAFWNS, YS, YV and YR were 91.44 ± 2.20%, 21.17 ± 1.57%, 30.01 ± 2.48%, 12.74 ± 1, 17%, 6.71 ± 1.21%, 7.61% ± 2.23, 20.11 ± 3.01%, respectively. The correlations of weight of fish yields were trivial, thus, they showed that the final weight is not a reliable indicator of yield cuts. The principal component analysis (PC) indicated that four of those tests were sufficient to explain 90% of the total variation, resulting from the combination of the variables final weight and yields. The performance evaluation of income on percentage of participation in the fish total weight of the different components of carcass depending on the final weight of the animals showed linear manner for (YEFWH, YH and YV). Surubim yields processing does not depend on stocking density and they have similar values.