Despite the severity of the prohibition of lashon ha-ra, an extensive system of specific laws and guidelines are not codified within the gemara or the Shulchan Arukh. This condition is understandable in light of the somewhat ambiguous nature of the prohibition. Though malicious attacks are forbidden, regardless of context, many of the more subtle forms of lashon ha-ra are highly dependent upon the framework, people involved, and manner of communication. The vacuum of information was largely filled by Rabbi Yisrael Meir Ha-kohen, who is better known by the title of his seminal work on the laws of lashon ha-ra - Chafetz Chaim.
Despite the absence of extensive details or perameters, there are two gemarot which offer general guidelines which help formulate the boundaries and define the essence of this prohibition.
The gemara in Bava Batra (39) declares, "Anything which is spoken in front of three people no longer poses an issue of lashon ha-ra" (the term employed by the gemara is "lishna bisha" - Aramaic for lashon ha-ra). The simple reading of this gemara (and the one adopted more or less by the Rambam in Hilkhot De'ot 7:5 and the Rashbam in Bava Batra) is that once a statement has been issued in the presence of three people, it is considered public knowledge. Assumedly, these people will disseminate the information, and it will become a known fact. Hence there is nothing wrong with retelling the gossip to others since they will inevitably discover it on their own. If this scenario, on a practical level, will not eventualize, even though three people have already heard the statement, it would be forbidden to retell this information. If the three who heard were sworn to secrecy, for instance (and we anticipate this oath being fulfilled) others are not allowed to repeat the information. Admittedly, a person would not be allowed to inform those who would not necessarily hear (those who live overseas, don't have e-mail, etc.) Likewise, the leniency only applies after someone has already told three people and violated the prohibition. Those three who heard (and presumably those who were subsequently informed by one of these three) would be allowed to pass on the information. To be sure, there are subtle differences in the positions of the Rambam and the Rashbam, but fundamentally they each subscribe to this reading and to this definition of lashon ha-ra.
A decidedly different view emerges from the Rabbenu Yona. He claims that outright malicious statements may not be repeated even if they have already been heard by three people (and will soon be disseminated to the broader public). The gemara's allowance applies only to ambiguous statements which can be read as either constructive or as denigratory. According to the Rabbenu Yona, if they are spoken in public, we can assume that the original intent was constructive or helpful. We might assume they were issued to warn others to avoid that individual or perhaps with the hope that the statement will ultimately reach the subject and help him reform or improve. On the other hand, words which are spoken in private and can have negative connotations are generally intended to be harmful. In any case, this whole discussion applies only to nebulous statements which support multiple meanings or intentions. Outright slander - even if true - would be forbidden in any context.
Possibly the Rambam and the Rabbenu Yona debate the essence of the prohibition of lashon ha-ra. According to the former, lashon ha-ra is forbidden because of the damage which it will cause the victim - whether monetary, psychological or even to his reputation. Once the damage is inevitable, no prohibition applies. According to the Rabbenu Yona, however, lashon ha-ra is inherently forbidden as a devious or underhanded act. Even if I don't augment the damage, I have acted in a duplicitous and forbidden manner. Thus the presence of three people only helps us decide whether an ambiguous statement carries malicious or benign (and possibly helpful) intent. Any clearly malicious statement, though, is categorically prohibited, regardless of whether its issuance will increase damage or merely restate information which others will soon discover on their own.
A third manner of reading the gemara in Bava Batra emerges from Tosafot. This reading views lashon ha-ra in a manner similar to the Rabbenu Yona. The gemara in Eruchin (15b) cites Rebbi Yossi, who claims that he never spoke in a manner which forced him to look over his shoulder and see who was listening. The simple reading of this statement is that Rebbi Yossi was attesting to his own piety in that he never said anything harmful and never had to be concerned who his audience. As Rabba deduced, anything which is stated in front of the subject is clearly not lashon ha-ra. If it is stated in front of him, it is obviously not injurious. Tosafot in Bava Batra read this gemara differently: Anything said in front of the subject – even if slanderous or insidious - is not considered lashon ha-ra. A person might have violated other infractions (embarrassing someone in public, ona'at devarim etc.) but has not committed lashon ha-ra. Tosafot extend this principle to explain the leniency of saying something in front of three people. Since whatever is said in front of three people will ultimately reach the ears of the subject, we consider it as if it were stated literally in his presence and thus not a violation of lashon ha-ra. Tosafot maintain that as malicious as a statement might be, it cannot be considered lashon ha-ra until it is devious or deceitful. When stated in an open manner in the presence of the subject, or even when stated in front of three people, in which case the speaker assumes that his subject will discover the comments, there is no violation of lashon ha-ra. Obviously, Tosafot saw lashon ha-ra as prohibited for the nature of the act and not the quotient of damage which ensues. As long as the act isn't duplicitous but open and honest, even though it is injurious, it is not considered lashon ha-ra.
An interesting gemara in Sanhedrin (31a) might shed light upon our question. The gemara demands that after issuing a verdict, judges should not reveal their differing opinions but should present a united front. A dissenting judge should not inform the losing party that he indeed ruled in that party's favor but was outvoted. As this would create damage to the reputations of the other judges, it is considered rechilut (the term used in Vayikra 19:16 and a cousin of lashon ha-ra – see afterword). The gemara extrapolates from this ruling the prohibition of revealing secrets. It even cites a pasuk in Mishlei (11:13), which equates a "holekh rakhil" (a talebearer) with someone who reveals secrets. The gemara cites by way of example a student who revealed certain secrets of the beit midrash twenty-two years after hearing them and was promptly evicted from the beit midrash by Rav Ami. Would this not indicate that lashon ha-ra is forbidden because the basic act is unethical? Assumedly, in the case of revealing "secrets" of the beit midrash, no harm was caused to any particular individual. However, since certain discussions were meant to be concealed, the revealer of secrets has betrayed his word and disrespected his peers!!
The Torah does not mention the prohibition of lashon ha-ra explicitly. It is derived from the related prohibition of "Lo teileikh rakhil be-amekha" (you should not bear tales in your society Vayikra 19:16). The Yerushalmi in Pe'ah (1:1) splits this pasuk into two different prohibitions - rechilut and lashon ha-ra. The Rambam in Hilkhot De'ot (perek 7) outlines the differences between these two prohibitions. Possibly, their difference might lie along the lines discussed earlier. One prohibition might be based upon the resulting damage to the victim, while the other might revolve around the immoral act itself. It should be noted that the gemara in Sanhedrin (31a) which spoke about revealing secrets equated this conduct to rechilut, NOT lashon ha-ra. Assuming the two are different, we cannot prove the nature of lashon ha-ra from a gemara which discusses rechilut.