Late Cenozoic andesitic rocks of the Colombian Andes are characterized by small but systematic and well-correlated variations in 87Sr/86Sr, 143Nd/144Nd, and δ18O that reflect significant crustal contamination. The range in isotopic ratios measured in northern Andean andesites is most pronounced for oxygen, relatively much more subtle for Sr and Nd. Pb isotopic data, so diagnostic of crustal contamination in central Andean lavas, exhibit minimal variation in the northern Andean volcanics. Trace element trends and phenocryst populations suggest dominantly mafic mineral fractionation, although plagioclase fractionation may be significant in later-stage differentiation. An assimilation-fractional crystallization model is constructed to match measured isotopic data. The results of that modelling indicate that the andesitic magmas of Galeras and Ruiz volcanoes may have assimilated up to 10–20% of crustal material. Correlated variations in measured isotopic ratios can be modelled satisfactorily by assuming a crustal contaminant of composition δ18O = 10 per mil, 87Sr/86Sr = 0.710–0.712 (Sr ≈ 90ppm), and 143Nd/144Nd=0.5120–0.5122 (Nd ≈ 20–30 ppm). We estimate that the composition of the primary magma (prior to crustal contamination) was in the range δ18O = 6.5 per mil, 87Sr/86Sr = 0.704, and 143Nd/144Nd = 0.51285, values that may suggest an additional crustal component was present in the primary magma. If we assume the added crustal component was derived from partial melting of subducted continental sedimentary material and/or oceanic crust (source contamination), then model calculations suggest that crustal material comprises less than 10 to 15 wt% of the primary magma. This model-dependent effect of source contamination is to shift isotopic ratios to the right on the Sr–Nd plot (see also Hawkesworth et al. 1977), whereas crustal contamination of the magma during ascent through the Andean crust appears to have produced isotopic trajectories that cut downward and through the mantle array.