Weather-dependent renewable generation and demand are key inputs to long-term energy-system planning. Computational constraints, however, often require capacity-expansion models to represent meteorological conditions using a single historical weather year or a reduced temporal dataset. Such representations implicitly assume that the selected period captures the relevant characteristics of the long-term climate. This study examines this assumption using 39 years of pan-European data for onshore wind, solar PV, reservoir hydropower inflow, and space-heating demand. Individual weather years are evaluated against the long-term record in terms of annual conditions, seasonal characteristics and cross-variable correlations. Results show that interannual variability differs considerably across weather-dependent inputs, with the largest and most regionally heterogeneous variations observed for reservoir hydropower inflow, while heat demand and wind show relatively less variability and solar PV generally exhibits the narrowest distributions. Seasonal characteristics and correlation structures are generally more consistent between years. Crucially, representativeness across these characteristics does not coincide: a year that closely represents the long-term record in one dimension may deviate substantially in another. Datasets used for long-term energy-system planning should be assessed against relevant annual, spatial, seasonal and cross-variable characteristics of the long-term record. The assessment framework developed here provides a means of evaluating whether such datasets retain the characteristics required for long-term energy-system planning.