Wake-flow data of the Nysted offshore wind farm is analysed with the engineering wake model of Jensen. Two-parameter fits referring to wake-decay parameter and wind direction are made to each individual 10 min-averaged recording of turbine powers. The fitted wind direction neither agrees with the measurements at the meteorological masts nor the turbine orientations, but is in agreement with the power deficit ratios between second-and first-row turbines. The fitted wake-decay parameter varies roughly between k = 0.02 and 0.04, depending on the wind direction with respect to the wind farm geometry. Its average ≈ 0.03 is smaller than the current standard k = 0.04, indicating that offshore wakes might be more narrow. No significant dependence on flow stability is found. Compared to two-parameter fits, one-parameter fits with pre-assigned wind direction perform worse; they lead to ≈ 0.04 and tend to overpredict wind farm power.