Karnik-Mendel algorithm involves execution of two independent procedures for computing the centroid of an interval type-2 fuzzy set: the first one for computing the left endpoint of the interval centroid (which is denoted by c l ), and the second one for computing its right counterpart (which is denoted by c r ). Convergence of the discrete version of the algorithm to compute the centroid is known, whereas convergence of the continuous version may exhibit some issues. This paper shows that the calculation of c l and c r are really the same problem on the discrete version, and also we describe some problems related with the convergence of the centroid on its continuous version.