Human-provided nesting shelters such as nest-boxes mitigate the shortage of natural breeding sites. Since artificial nests are not where animals evolved and optimised their reproductive performance, it remains inconclusive if these are adequate substitutes, ensuring equivalent fitness returns while breeding. In particular, most knowledge on the ecology of cavity-nesting birds comes from nest-box populations, but no study has directly compared fitness consequences of breeding inside nest-boxes in relation to natural cavities in cities. We directly compare the reproductive performance, life-history trait variation and fitness consequences for two small passerines, blue and great tits, breeding in nest-boxes as opposed to natural cavities in an urban deciduous forest. We use a quasi-experimental setting to comprehend the conservation potential of these artificial cavities and to support/question generalisations stemming from nest-box studies. We show that the effects of cavity type vary between species: in blue tits, fitness proxies were negatively affected by nest-boxes (lower fledging success and fledgling numbers, longer time spent in the nest and later fledging date relative to natural cavities), while in great tits, the fitness proxies were unaffected by cavity type. Importantly, we detected accelerated incubation in both species breeding in nest-boxes. No differences in pre-hatching traits (lay date, clutch size, hatching rates) between cavity types suggest that the fitness deterioration occurred because of post-hatching effects. We highlight the ecological importance of old-growth tree stands, providing natural cavities for city-breeding animals and the need for quantifying alterations of reproductive ecology in other taxa using human-provided nests. Due to the detected cavity type-dependent variation in reproductive performance, we support the criticism regarding the unconditional extrapolation of evolutionary and ecological interpretations of nest-box studies to general populations.