When auditing for CRA and reviewing a large sample of loans to evaluate the bank's level of lending to borrowers of different income levels, I have always used Median Family Income for analyzing residential real estate loans and Median Household Income for non-real estate consumer lending. Is this still considered the correct way to do it?

The FFIEC EMFI for the MSA or statewide non-MSA is the correct denominator to determine the income class of mortgage borrowers. Be sure to use the EMFI that pertains to the year the loan originated.

There's no official "correct" formula for consumer loans, but using the estimated median household income for the MSA or statewide non-MSA for the year the loan was originated would be consistent.

Cindy, to add just another dimension - As you know, CRA in income driven, LMI tracts and individuals, while ECOA is race, gender and ethnicity driven.
During a fair lending exam, regulators take guidance from ECOA and FHAct for discrimination, and the primary driving points are minority individuals and tracts, not income. You may want to additionally identify the percentage of applications and loans compared to minority population in your assessment area, and additionally to see if the market penetration to Majority-Minority tracts (in terms of loans) is similar to the overall market penetration in your AA.