The short-term statistics of wave conditions in coastal waters around the UK have been investigated using over 40,000 half-hour long sea state records with significant wave height greater than 3 m. The extensive data set facilitates an assessment of various wave height and period distribution models in shallow and intermediate waters. The results reveal that the relative wave height H_s/D (where H_s is the significant wave height and D the water depth) can serve as a key indicator in choosing the distribution with least error in a given sea state. The Naess model is found to be the most accurate in describing the tail of the wave height distribution in a sea state for low relative wave heights (H_s/D < 0.2), and the depth-dependent van Vledder model for high relative heights (H_s/D > 0.4). In between these sea states, a transition in the performance of the deep-water and the depth-dependent models is visible. When details of the spectrum are not available, the Weibull distribution is the most accurate amongst applicable models, in spite of considerable variability in its parameter values. While the spectral bandwidth appears to have minimal impact on the distribution of wave heights in a sea state, it does appear to influence the distribution of wave periods. Wave period relationships based on measurements are found to deviate from empirical relationships proposed, for example, by the US Army Coastal Engineering Manual. Improved formulas that incorporate the spectral width are therefore proposed.