Background Policy makers in England advocate referral of patients with suspected dementia to Memory Assessment Services (MAS), but it is unclear how any improvement in patients' health-related quality of life (HRQL) compares with the associated costs. Aims To evaluate the cost-effectiveness of MAS for the diagnosis and follow-up care of patients with suspected dementia. Method We analysed observational data from 1318 patients referred to 69 MAS, and their lay carers (n = 944), who completed resource use and HRQL questionnaires at baseline, three and six months. We reported mean differences in HRQL (disease-specific DEMQOL and generic EQ-5D-3L), quality-adjusted life years (QALYs) and costs between baseline and six months after referral to MAS. We also assessed the cost-effectiveness of MAS across different patient subgroups and clinic characteristics. Results Referral to MAS was associated with gains in DEMQOL (mean gain: 3.48, 95% confidence interval: 2.84 to 4.12), EQ-5D-3L (0.023, 0.008 to 0.038) and QALYs (0.006, 0.002 to 0.01). Mean total cost over six months, assuming a societal perspective, was £1899 (£1277 to £2539). This yielded a negative incremental net monetary benefit of -£1724 (-£2388 to -£1085), assuming NICE's recommended willingness-to-pay threshold (£30,000 per QALY). These base case results were relatively robust to alternative assumptions about costs and HRQL. There was some evidence that patients aged 80 or older benefitted more from referral to MAS (p < 0.01 from adjusted mean differences in net benefits) compared to younger patients. MAS with over 75 new patients a month or cost per patient less than £2500 over six months were relatively more cost-effective (p < 0.01) than MAS with fewer new monthly patients or higher cost per patient. Conclusions Diagnosis, treatment and follow-up care provided by MAS to patients with suspected dementia appears to be effective, but not cost-effective, in the six months after diagnosis. Longer term evidence is required before drawing conclusions about the cost-effectiveness of MAS.