Proteomic profiling by MALDI-TOF MS presents various advantages (speed of analysis, ease of use, relatively low cost, sensitivity, tolerance against detergents and contaminants, and possibility of automation) and is being currently used in many applications (e.g. peptide/protein identification and quantification, biomarker discovery, and imaging MS). Earlier studies by many groups indicated that moderate reproducibility in relative peptide quantification is a major limitation of MALDI-TOF MS. In the present work, we examined and demonstrate a clear effect, in cases apparently random, of sample dilution in complex samples (urine) on the relative quantification of peptides by MALDI-TOF MS. Results indicate that in urine relative abundance of peptides cannot be assessed with confidence based on a single MALDI-TOF MS spectrum. To account for this issue, we developed and propose a novel method of determining the relative abundance of peptides, taking into account that peptides have individual linear quantification ranges in relation to sample dilution. We developed an algorithm that calculates the range of dilutions at which each peptide responds in a linear manner and normalizes the received peptide intensity values accordingly. This concept was successfully applied to a set of urine samples from patients diagnosed with diabetes presenting normoalbuminuria (controls) and macroalbuminuria (cases).