We analysed the patterns of variation that characterize 33 catch time series of large pelagic fishes exploited by the Japanese and Taiwanese longline fisheries in the Indian Ocean from 1968 to 2003. We selected four species, the yellowfin (Thunnus albacares), the bigeye (T. obesus), the albacore (T. alalunga), and the swordfish (Xiphias gladius) and aggregated data into five biogeographic provinces of Longhurst (2001). We carried out wavelet analyses, an efficient method to study non-stationary time series, in order to get the time-scale patterns of each signals. We then compared and grouped the different wavelet spectra using a multivariate analysis to identify the factors (species, province or fleet) that may influence their clustering. We also investigated the associations between catch time series and a large-scale climatic index, the Dipole Mode Index (DMI), using cross wavelet analyses. Our results evidenced that the geographical province is more important than the species level when analyzing the 33 catch time series in the tropical Indian Ocean. The DMI further impacted the variability of tuna and swordfish catch time series at several periodic bands and at different temporal locations, and we demonstrated that the geographic locations modulated its impact. We discussed the consistency of time series fluctuations that reflect embedded information and complex interactions between biological processes, fishing strategies and environmental variability at different scales.