Democrats said yesterday that they've found 432 felons who appear to have voted illegally in November but were ignored by Republicans in their legal challenge to the governor's election.

Republicans left those felons out of their case, Democrats say, because the illegal votes came from parts of the state that supported Republican Dino Rossi in his race against Christine Gregoire, boosting the chances that they were in Rossi's favor.

In a court filing yesterday, Democrats also outlined other alleged illegal votes and errors by election officials that they say favored Rossi in the election and more than make up for any similar charges by Republicans.

"They essentially cherry-picked through Democratic portions of the state to come up with their list," state Democratic Party Chairman Paul Berendt said yesterday.

Mary Lane, a spokeswoman for Rossi, denied that Republicans' list of 946 felon voters targeted Gregoire territory.

"Obviously it's a laborious process," she said. "But we looked everywhere."

Rossi lost the election to Gregoire by 129 votes after a hand recount reversed the outcome of the original count and a machine recount.

Even before yesterday's announcement, Republicans had moved to stop Democrats from introducing evidence of illegal votes or election errors that might offset evidence collected by Republicans.

Chelan County Superior Court Judge John Bridges will hear arguments Monday on whether the Democratic evidence can be considered in the Republican lawsuit asking that the election be overturned and Gregoire removed from office. The trial is set to begin May 23.

Berendt said the Democratic list of alleged felon voters will grow before Friday's deadline for the party's attorneys to file allegations of illegal votes.

"Their story keeps changing," she said. "First they said our list was horrible, inaccurate and sloppy. Then they were saying most felons vote Republican anyway. Now it looks like they're playing catch-up."

Democrats looked for felon voters in counties that Rossi won in the November election and in Rossi-leaning pockets of counties that Gregoire won.

Alleged felon voters were found in 31 counties, according to the Democrats' court filing yesterday, including 65 in Spokane, 55 in Yakima and 35 in Clark. In King County, Democrats say, they found 31 felons from Rossi strongholds.

Searching for felons has proved to be fraught with problems for the political parties, their supporters and the media. Records are often unclear and require follow-up to eliminate people included by mistaken identity.

To press their case, Democrats are replicating Republican moves they've criticized and that they argue in court should be disallowed.

As evidence that a felon voted illegally, Democrats used county voter-crediting information that, when used by Republicans, they said was "inherently unreliable to prove that a ballot was given to an individual, marked by the individual, cast by the individual, and counted by election officials."

Democrats have also criticized Republicans for naming alleged felons without fully vetting the names, saying the exposure violates "common decency."

But yesterday Democrats filed court papers that included a list of 432 people they say "appear to be felons who voted" though the party's investigation is continuing.

And to show that their felons offset voting by Republicans' felons, Democrats will rely on a methodology that they say should be disallowed in the court case.

Republicans have proposed using a statistical analysis to apportion illegal votes by the same percentage as legal votes. For example, if Gregoire won 60 percent of the votes in a given precinct, Republicans say the court should assume that 60 percent of any illegal votes in that precinct were cast for her.

Two academic experts hired by Republicans say their calculations using that method show Rossi would win.

In court papers, Democrats call that "speculative attribution" and "based on chance." Berendt yesterday called it "guessing."

But he said Democrats will use the same formula to show statewide votes by felons did more to help Rossi than to hurt him.

Berendt said Democrats still will oppose the Republican methodology, but also are ready to counter if it's allowed.

"We believe that even if the judge were to accept the theory, there is no proof whatsoever that Dino Rossi won the election," Berendt said.

The Democrats also said they found 1,939 improperly counted provisional ballots in counties other than King. All but three of those counties went for Rossi in the manual recount.

Counties with the largest numbers of mishandled ballots cited by Democrats were Whitman (783), Stevens (400), Walla Walla (331), Pierce (164) and Adams (117).

Provisional ballots are cast by voters whose names don't appear in the poll book or who vote away from their regular polling places. State regulations don't allow those ballots to be counted until voters' registrations and signatures have been verified.

Republicans have cited 785 mishandled provisional ballots in King County, and of those, 122 were cast by people not properly registered.

While Democrats say their investigation is continuing, they don't specify how many of the provisional ballots on their list were actually cast by unregistered voters. It's possible that votes were valid even if they didn't go through proper checks.

In Whitman County, Auditor Eunice Coker conceded that workers did not follow proper procedures but said they were still able to determine the ballots were cast by registered voters.

King County elections director Dean Logan has said the mishandled provisional ballots concern him more than any other election error.

In a deposition last week, Logan's deputy, Bill Huennekens, said some precinct inspectors who worked on Election Day will not work future elections because of the mishandled provisional ballots. He did not know how many would not be asked back.

Times staff reporter Keith Ervin contributed to this report. David Postman: 360-943-9882 or dpostman@seattletimes.com