We frequently get questions from patrons who are interested in finding demographic data for neighborhoods in New York. Finding this data can be tough, as the census does not issue data for neighborhoods - you have to take the levels of geography they use (census tracts, zip codes) and force it to fit neighborhood boundaries. The NYC Department of Planning has done this and they've issued profiles for each neighborhood / community district in the city.
The problem is that this data is from the 2000 Census and is fairly old. The Census Bureau does provide recent data through the American Community Survey (ACS), but the data is limited to large geographic areas. However, the ACS has begun to release three year averaged estimates for areas called PUMAs, which are aggregations of census tracts. And the NYC Dept of Planning has done it again - they've created a whole series of profiles for each community district within the city based on PUMAs.
The 2005-07 ACS PUMA data are currently the most recent census figures at the smallest geographic level possible within the city, and are the closest thing we have to neighborhood data.
The profiles are available in pdf format on the NYC Dept of Planning's ACS page, and are listed under the Community Districts column. The data includes demographic, social, economic, and housing profiles. Each pdf file contains profiles for every single community district, sorted by borough and community district number. This isn't exactly ideal, as you have to do a lot of scrolling to find the district you want. Just above the table is a link to a busy map of PUMAs and Community Districts (the two areas don't line up exactly). Use the map to identify the part of the city you're interested in, note what the community district number is, and then open up a report and page through by borough and district number. This may not seem ideal, but is a million times easier than trying to build your own neighborhood geography and profiles using the American Factfinder.
Next year, the Census will update the averages by dropping the oldest year and adding the newest one (so we'll have 2006-2008 estimates). The estimates are three-year averages rather than annual numbers due to the small sample size. The estimates have a confidence level of 90% and each number has a margin of error associated with it. NYC Planning provides an overview (in pdf at the top of the profile page) of what the ACS is and they explain how and why they use it to create the profiles.