It is sometimes assumed that changes in risks of coastal flooding can be computed by just adding mean sea level changes to existing statistics of tidal and meteorological effects. This assumption can be examined by looking at trends in separate tidal and nontidal sea levels, but suitable sites with long records of good quality sea level measurements are rare. As an example, hourly sea level data from the well-maintained tide gauge at Newlyn in southwest England has been critically edited for errors and separated into tidal, nontidal, and mean sea level components. These components have then been analysed for significant trends, and possible correlations with meteorological indicators. There is a small (0.16 mm y⁻¹) increase in standard deviation of the observations, and consistent increases in the M₂ and M₄ tidal amplitudes. An apparent reduction in the standard deviation of the nontidal residuals can be attributed to a change to a different measuring system in 1984, as can a sudden decrease of 2° (4 min earlier) in the phase lag of M₂. There are weak correlations between nontidal (surge) statistics and meteorological parameters, including the North Atlantic oscillation (NAO). Observed annual maximum sea levels are increasing at a rate not significantly different from the observed increase in mean sea level of 1.77 ± 0.12 mm y⁻¹. These very exact analyses of a high-quality consistent long sea level series may indicate limits on the types of trends to be expected elsewhere in the region. Application of this type of analysis for estimating future flooding risks is discussed.