Consistent individual differences (CIDs) in behaviour, indicative of behavioural types or personalities, have been shown in taxa ranging from Cnidaria to Mammalia. However, despite numerous theoretical explanations there remains limited empirical evidence for selective mechanisms that maintain such variation within natural populations. We examined behavioural types and fitness proxies in wild female grey seals at the North Rona breeding colony. Experiments in 2009 and 2010 employed a remotely-controlled vehicle to deliver a novel auditory stimulus to females to elicit changes in pup-checking behaviour. Mothers tested twice during lactation exhibited highly repeatable individual pup-checking rates within and across breeding seasons. Observations of undisturbed mothers (i.e. experiencing no disturbance from conspecifics or experimental test) also revealed CIDs in pup-checking behaviour. However, there was no correlation between an individuals' pup-checking rate during undisturbed observations with the rate in response to the auditory test, indicating plasticity across situations. The extent to which individuals changed rates of pup-checking from undisturbed to disturbed conditions revealed a continuum of behavioural types from proactive females, who maintained a similar rate throughout, to reactive females, who increased pup-checking markedly in response to the test. Variation in maternal expenditure (daily mass loss rate) was greater among more reactive mothers than proactive mothers. Consequently pups of more reactive mothers had more varied growth rates centred around the long-term population mean. These patterns could not be accounted for by other measured covariates as behavioural type was unrelated to a mother's prior experience, degree of inter-annual site fidelity, physical characteristics of their pupping habitat, pup sex or pup activity. These findings are consistent with the hypothesis that variation in behavioural types is maintained by spatial and temporal environmental variation combined with limits to phenotype-environment matching.