Several methods currently exist to quantitatively reconstruct palaeoclimatic variables from fossil botanical data. Of these, probability density function (PDF)-based methods have proven valuable as they can be applied to a wide range of plant assemblages. Most commonly applied to fossil pollen data, their performance, however, can be limited by the taxonomic resolution of the pollen data, as many species may belong to a given pollen type. Consequently, the climate information associated with different species cannot always be precisely identified, resulting in less-accurate reconstructions. This can become particularly problematic in regions of high biodiversity. In this paper, we propose a novel PDF-based method that takes into account the different climatic requirements of each species constituting the broader pollen type. PDFs are fitted in two successive steps, with parametric PDFs fitted first for each species and then a combination of those individual species PDFs into a broader single PDF to represent the pollen type as a unit. A climate value for the pollen assemblage is estimated from the likelihood function obtained after the multiplication of the pollen-type PDFs, with each being weighted according to its pollen percentage. To test its performance, we have applied the method to southern Africa as a regional case study and reconstructed a suite of climatic variables (e.g. winter and summer temperature and precipitation, mean annual aridity, rainfall sea-sonality). The reconstructions are shown to be accurate for both temperature and precipitation. Predictable exceptions were areas that experience conditions at the extremes of the regional climatic spectra. Importantly, the accuracy of the reconstructed values is independent of the vegetation type where the method is applied or the number of species used. The method used in this study is publicly available in a software package entitled CREST (Climate REconstruction SofTware) and will provide the opportunity to reconstruct quantitative estimates of climatic variables even in areas with high geographical and botanical diversity.