Real-world　negotiations　are　characterized　by　complex　negotiation　spaces,　tough　deadlines,　bounded　agent　rationality,　very　limited　information　about　the　opponents,　and　volatile　negotiator　preferences.　Classical　negotiation　models　fail　to　address　most　of　these　issues.　Practical　negotiation　agents　with　an　effective　and　efficient　fuzzy　inference　to　deal　with　complex　and　incomplete　negotiation　spaces　arising　in　real-world　applications　are　proposed.　The　agent　with　the　fuzzy　inference　determines　the　values　of　the　new　offer　through　the　set　of　fuzzy　rules.　An　evolutionary　algorithm　with　Bayesian　learning　of　its　opponents＇　preferences　according　to　the　history　of　the　counter　offers　and　genetic　algorithms　(GA)　are　used　to　optimize　the　parameters　of　the　fuzzy　rules.　Simulation　shows　that　responsive　and　adaptive　negotiation　agents　work　for　real-world　negotiations.