Background: To ensure carers of people with dementia receive support, community services increasingly use measures of caregiver (carer) burden to assess for unmet need. This study used Bradshaw's taxonomy of need to explore the link between measures of carer burden (normative need), service use (expressed need), and carer's stated need (felt need). Methods: This mixed method exploratory study compared measures of carer burden with community services received and unmet needs, for 20 community-dwelling carer/care-recipient pairs. Results: A simple one-item measure of carers' felt need for more services was significantly related to carer stress as measured on the GHQ-30. Qualitative data showed that there are many potential stressors for carers, other than those related to the care-giving role. We found a statistically significant rank correlation (p = 0.01) between carer's use of inhome respite and the care-recipient's cognitive and functional status which is likely to have been related to increased requirement for carer vigilance, effort and the isolation of spouse carers. Otherwise, there were no statistically significant relationships between carer burden or stress and level of service provision. Conclusion: When carers are stressed or depressed, they can recognise that they would like more help from services, even if measures of carer burden and care recipient status do not clearly indicate unmet service needs. A question designed to elicit carer' felt need may be a better indicator of service need, and a red flag for recognising growing stress in carers of people with dementia. Assessment of service needs should recognise the fallibility of carer burden measures, given that carer stress may not only come from caring for someone with dementia, but can be significantly compounded by other life situations.