Abstract: The requirement of tide or current data length of orthogonal response analysis is researched through theoretical deduction and experiment.The results show that it requires at least 2 day-length of ideal data.But for the actual observations,it requires at least 7 day-length data in the diurnal tide dominated area,and 10 days in the semi-diurnal tide dominated area.Towards a shorter term data than described above,an improved model of orthogonal response analysis is found and used to analysis this type of data through introducing ratio of parameters and changing the value of Δτ.The result of experiment using 366 groups of data at 12 stations in China,which contains 1 day-length data in each group,showed that the parameters calculated from short term data are close to the ones from long term data(annual data).Finally,a comparison between orthogonal response analysis and quasi-harmonic analysis for short term data is studied through both theoretical deduction and experiment.The results showed that the former one has a higher precision and does not have to take good astronomic date into account.