Global land and ocean precipitation analyses are available using satellite data beginning 1979. Prior to 1979 gauge data provide coverage for much of the land, but over oceanic regions there are only a few island gauges. In recent years attempts have been made to reconstruct large-scale oceanic precipitation variations using the sparse gauge network and large-scale spatial covariance. Covariance statistics are based on analyses using satellite data. Reconstructions may begin 1950 or earlier. Earlier reconstruction efforts have resolved ENSO variations, which have large scales and strong signals. Recent studies at ESSIC/CICS have improved these reconstructions by using a new, more homogeneous satellite analysis for statistics and improved tuning of the method. This has improved reconstruction skill of interannual variations over the Northern Hemisphere extra-tropical oceans, and allowed reconstructions to be extended to 1900. However, the gauge-based methods do a poor job at oceanic inter-decadal variations for the satellite period. To address that problem, a new method was developed using a CCA to compute the annual-average precipitation anomaly from annual-averages of sea-level pressure and sea-surface temperature anomalies. Initial results suggest that this method resolves the large-scale inter-decadal variations. At present we are developing a two-part reconstruction, using the CCA-based method to analyze inter-decadal variations and the gauge-based analysis to resolve interannual variations. An improved, longer homogeneous satellite-period analysis is also under development to provide improved reconstruction statistics.