Provided are methods, computer programs and data processing apparatus using numeric processing. Firstly, a corrective mechanism enables a method for performing accurate integer divisions to be derived from an approximate division method which does not, of itself, always produce an accurate result but for which the range of errors is known. By applying the corrective mechanism to a suitable approximate division method, a numeric processor or software-implemented numeric processing mechanism implementing the invention can perform the integer division operation efficiently. Secondly, an approximate division method which uses only rapid operations for fast integer division, and which has only a small possible range of errors, is used to enable the correction method to be completed rapidly. This addresses problems encountered when attempting to apply known methods to the task of performing integer division by large divisors or in a limited-size numeric register, and can provide efficiency improvements for a wide range of data processing systems and applications of those systems.