This paper reviews some of the design decisions affecting the quality of various forms of estimates produced from SIPP data.  Many of the SIPP design alternatives are also applicable in other surveys, especially other panel surveys.  Nonsampling errors which arise from the major SIPP decision to employ a panel design, namely panel nonresponse and panel conditioning, are discussed.  Other design decisions, such as the length of reference period, respondent rules, following rules, and data collection mode, are reviewed in light of their possible effects on nonsampling errors.  Finally, the measurement of gross change in a panel survey is discussed.  Examples using SIPP data are provided to clarify the discussion.