In order to document inter-individual variability in salivary protein patterns, unstimulated whole saliva was obtained from 12 subjects at 10 am and 3 pm of the same day. Saliva proteins were separated using two-dimensional gel electrophoresis, and semi-quantified using image analysis. One-way ANOVA was used to test the effects "time of sampling" and "subject". Data were further explored by multivariate analyses (PCA, hierarchical clustering). Spots of interest were identified by mass spectrometry (MALDI-TOF MS/MS and nanoLC ESI-IT MS/MS). A dataset of 509 spots matched in all gels was obtained. There was no diurnal statistical effect on salivary patterns while inter-individual variability was high with 47 spots differentially expressed between subjects (p<1%). Clustering of these spots revealed that subjects could be discriminated first based on several proteins participating to the non-specific immune response (cystatins, lipocalin 1, parotid-secretory protein and prolactin-induced protein). Independently, subjects were also differentiated by their level of proteins originating from serum and involved in the immune system (complement C3, transferrin, IgG2), as well as the relative abundance of enzymes involved in carbohydrates metabolism (amylase and glycolytic enzymes). Inter-individual variability should be accounted for when searching for salivary biomarkers or when studying in-mouth biochemical mechanisms.