Background The field of degree question is new to the 2009 American Community Survey (ACS) and the 2009 Puerto Rico Community Survey (PRCS). Two versions of the question were tested in the 2007 ACS Content Test, an open-ended version and a categorical version. The open-ended version performed better according to the evaluation criteria set forth for the content test (Raglin et. al., 2008). The purpose of this evaluation is to assess how the field of degree question is performing in production on both the ACS and the PRCS. On the 2009 ACS and PRCS English paper questionnaires, the field of degree question is located directly below the educational attainment question in the same column. However, on the paper questionnaires created for the 2007 ACS Content Test, the educational attainment question is at the bottom of one column while the field of degree question is at the top of the next column. The question was tested this way because the categorical version of the question was longer and would not fit under the educational attainment question. Since the ACS wanted to compare both questions in the same placement on the test, both questions were tested at the top of the next column. We believe that this difference could cause differences in item nonresponse between the English versions of the questionnaires in the 2007 ACS Content Test and the 2009 ACS production. On the 2009 ACS and PRCS Spanish paper questionnaires, the educational attainment question is also at the bottom of one column while the field of degree question is at the top of the next column because of space constraints. Conclusions The new field of degree question seems to be working as expected in the 2009 ACS and PRCS. The distribution of responses is similar to that found in the 2007 ACS Content Test. The 2009 ACS HU outperformed the 2007 ACS Content Test in terms of item nonresponse rate and consistency, which is most likely because of the new placement of the question. Also, the new coding system is working very well for the ACS and the PRCS. However, the high percentage of uncodable answers coming from the CATI and CAPI modes of data collection should be investigated in more detail.