Understanding the short- and long-term dynamics of drug and multi-drug resistance is important for public health. Yet, there are pervasive trends in resistance dynamics that have not been fully explained. Firstly, antibiotic sensitive and resistant strains coexist robustly, despite prolonged selection pressure from antibiotics. Secondly, resistance to different antibiotics tends to co-occur on the same strains, leading to high frequencies of multi-drug resistance (MDR). First, we present a model in which coexistence is maintained by variation in duration of carriage within the pathogen population (e.g. pneumococcal serotypes differing in duration of carriage) because the fitness effect of resistance depends on duration of carriage. Second, we show that this model is structurally similar to other plausible models of coexistence where the coexistence-maintaining mechanism is based on variation in the fitness benefit of resistance, and that models with this structure also give rise to high MDR frequencies, because resistance against all antibiotics is concentrated in the sub-populations where the fitness advantage gained from resistance is high. We find that predictions about patterns of resistance and multi-drug resistance from this model are qualitatively consistent with trends observed in multiple Streptococcus pneumoniae datasets. This model provides a parsimonious explanation for the pervasiveness of high MDR frequencies and allows us to reconcile this trend with observed long-term stability in the prevalence of resistance.