The article is devoted to the issues related with the quality appraisal of computer algorithms according to labour-intensiveness criteria. The traditional appraisal of an average labour-intensiveness makes it possible to get significant results only as a matter of statistics that is to appraise an algorithm as a multiple input unit with the fixed length. The article presents an interval appraisal - a reliable labour-intensiveness based as by analogy with reliable intervals of math’s statistics. It is presupposed to use beta-distribution to approximate the allocation of labour-intensiveness value as a limited accidental quantified value. A methodology to define the reliable labour-intensiveness as a function for algorithm input length is introduced as well.