"I find many of the terms of the settlement to be either confusingly vague, subject to manipulation or both," committee chairman Patrick Leahy said at a hearing yesterday.
Leahy also said that the proposed enforcement mechanism, which includes a three-member technical oversight committee that reports to the Justice Department, "lacks the power and the timeliness necessary to inspire confidence in its effectiveness".
Concerns about the agreement, reached last month by the US Department of Justice and nine of the 18 states in involved in this case, were bipartisan and they all sprang from the same source: the 1995 consent decree barring Microsoft from anticompetitive licensing practices. The software giant's current antitrust case springs in part from allegations that it violated that decree.
Senator Orrin Hatch also had doubts about the effectiveness of the proposed settlement, saying it "only appears to be the end of the latest chapter".
The Senate has no real authority in this case, and the committee was exercising its power of "oversight" on the Justice Department. The settlement is now undergoing a review by District Judge Colleen Kollar-Kotelly.
Nine other states, led by California, Connecticut and Iowa, are pursing litigation in search of tougher remedies.
Meanwhile, Microsoft filed its response to the proposed remedy from the nine states holding out on settlement, accusing them of acting in the interest of companies located in their backyards.
"It is apparent... that the nonsettling states seek to punish Microsoft and to advance the commerce interests of powerful corporate constituents - Microsoft competitors such as Sun, Oracle, Apple and Palm," the company said.
In its brief, Microsoft questioned whether the nine states even have the authority to recommend remedies "on behalf of the citizenry of the entire country".
The nine dissenting states are seeking broader restrictions, including a requirement that Microsoft make Internet Explorer code open-source, port Office to other operating systems and produce a thin client.