Still unhappy with a 1984 government estimate that America's homeless number only 250,000 to 350,000, a House subcommittee has scheduled new hearings on charges that the Department of Housing and Urban Development deliberately underestimated.

Rep. Henry B. Gonzalez (D-Tex.) asked Kathleen Peroff, a former HUD official with the Office of Management and Budget, who directed the 1984 study, to appear July 17 to clear up "discrepancies."

Gonzalez said in a letter that he wants her to explain how the study was done and to answer charges by several professors who examined HUD survey documents and by Mitch Snyder, an activist for the homeless, that the methodology was manipulated to produce a low total. Snyder is suing HUD to force it to repudiate the report. His suit was dismissed in district court but is being appealed.

Gonzalez, chairman of the House Banking subcommittee on housing, and others have made it clear that they are suspicious that the report greatly underestimated the homeless because the Reagan administration "refuses to acknowledge the extent of the national problem," Gonzalez said.

Gonzales said he thinks that the HUD count -- the government's official measure of how many people in this country are homeless -- is being used to justify a low level of federal assistance.

The HUD report surprised many advocates for the homeless because some sources had estimated that 2 million to 3 million people are homeless.

The major charge against the report -- as laid out in subcommittee documents and in a lengthy letter to several members by Timothy D. Junkin, Snyder's attorney -- is that HUD's total figure for the homeless in some metropolitan areas was actually only the total for the central city and perhaps one or two adjoining counties.

Witnesses at a December hearing conducted by Gonzalez said that HUD used as the basic geographic unit for estimating numbers of homeless a large area called the Ranally Metropolitan Area (named after map publisher Rand McNally, which devised the system for use in its annual atlas of market areas). The Ranally Area consisted of a central city, its immediate suburbs and numerous farther-out suburbs and related cities.

But they said that HUD's surveys in many cases estimated the homeless only in the central city and few near-in suburbs, then used that figure as the total of homeless in the entire Ranally area, resulting in gross underestimation of totals.

For example, according to testimony at a December hearing by professor Richard Applebaum of the University of California at Santa Barbara, the HUD surveys in the New York Ranally Area, with 16.6 million people, ignored 13 counties with 41 percent of the 9.5 million suburban population of the area.