This paper determines the market value of dividends in the UK during periods before and after 1997. Previous studies, which use the ex-dividend day method, tend to provide noisy and potentially biased measures of dividend value. We estimate the value of dividends from the prices of shares that are identical except for their dividend entitlements, and are traded concurrently (within the same hour). We argue that our estimates of dividend value are the cleanest yet available for the UK. Our evidence suggests that ex-dividend day estimates are biased downwards, but that this bias may be mitigated by the use of robust regression. Dividend values are heterogeneous and are not explained by the tax-clientele hypothesis.