I see you have some good articles to look over that may help. Most hydrology books discuss the topic. Make sure watersheds are in same physiographic area. Some stream types flood frequently such as braided streams, while others deeply entrenched as gullies have lost access to floodplain and seldom flood to its abandoned terrace level. Using several gaged streams is better than one. Even with many stations across a regionl physiographic area or state, expect some variability such as maybe plus or minus 40% as confidence limits.
If the ungauged (d/s) and gauged (u/s) stations are on the same river, you can use routing techniques to route the flood hydrograph from gauged to ungauged. You can perform a flood frequency analysis using the routed flood hydrographs at the ungauged station. Various textbook studies and empirical methods are already available for flood routing. This approach may produce results that are uncertain, so that should be considered as well.
Use Pasini's and Ventura's methods as well as watershed area ratio method, but You have to make sure that you are getting an accurate CN for the Watershed.