At first blush, it looked like the best news the Washington real estate market had gotten in months.

But when the National Association of Realtors (NAR) reported Monday that the number of home sales in the District of Columbia climbed by 6.2 percent in the third quarter compared to the same period last year, the main thing that rose were eyebrows.

"It just ain't so," said Rufus S. Lusk III, president of a Silver Spring-based real estate information company that tracks home sale transactions throughout the region. "D.C. is a basket case."

"Downright fraudulent and negligent" is how George Altoft, senior assessor for the District, described the numbers.

In actuality, he said, sales of single-family homes fell 19.3 percent during that period and condominium sales fell 26.2 percent. He questioned the motivation of the real estate trade group in presenting what he called "a false picture" of the market.

Yesterday, an NAR official acknowledged that the home sales figures the organization reported for the District were mistakenly skewed upwards by as much as 25 percent.

"It looks like we received some spurious data from the D.C. Association of Realtors," said Glenn Crellin, NAR's vice president for economics and research.

"The magnitude of the discrepancy bothers me a great deal."

Crellin said he believed that the D.C. association recently had shifted its multiple listing system procedures from reporting closed sales, or completed transactions, to reporting signed contracts, which may or may not result in actual sales. He said the problem was magnified when the national real estate group adjusted the numbers seasonally to compute an annualized sales figure.

He said he was sure the local real estate group had "no intent to mislead" people interested in the state of the city's residential real estate market.

For his part, the president of the local association, Don Denton of Dale Denton Real Estate Inc., said he stands behind the numbers that were provided to the NAR.

He also questioned the motivation of the national real estate group in disseminating inaccurate information if officials believed the numbers were inaccurate.

Crellin "shouldn't have released the numbers then," Denton said. "... Nobody is trying to hide anything."

The real problem, Denton said, is that different groups compile homesales statistics in different ways depending on their particular needs.

He said that the numbers compiled by groups such as Lusk and the D.C. assessor can be equally misleading because they include property transfers within families, mistakenly classify homes with small rental units as multifamily properties and fail to include co-op sales.

Clearly, however, the D.C. real estate association's tabulations included fewer transactions than Altoft tallied through the D.C. recorder of deeds. Altoft said that 1,502 single-family homes and condominiums were sold in the District in the third quarter of 1990, down 21.7 percent from the same period in 1989, when the assessor recorded 1,918 transactions. Denton said his group recorded about 1,100 sales of single-family homes and condominiums in the three-month period, a 6.2 percent increase from the sales his group recorded for the same period in 1989.

"The fact of the matter is there are no accurate numbers," Denton said.

"The numbers are a moving crap game. Their numbers aren't perfect and neither are ours."

Altoft and Lusk said the discrepancy is significant and troubling.

Altoft said that it is crucial that the numbers be accurately reported because District taxpayers should not be lulled into a false sense of complacency about the local tax base. The decline in sales translates into a significant loss in tax revenues for the District because transfer and recordation tax income, which total 2.2 percent of the purchase price of homes that change hands, will be reduced dramatically.

Lusk said accurate numbers are important because the market will recover only when sellers become more realistic about sales prices.

It is unclear whether a mistake concerning the District market could have skewed the regional median sales price of $153,400 reported by the NAR. Crellin said it might have had "a marginal impact."