Background: The brucellae are facultative intracellular bacteria that cause brucellosis, one of the major neglected zoonoses. In endemic areas, vaccination is the only effective way to control this disease. Brucella melitensis Rev 1 is a vaccine effective against the brucellosis of sheep and goat caused by B. melitensis, the commonest source of human infection. However, Rev 1 carries a smooth lipopolysaccharide with an O-polysaccharide that elicits antibodies interfering in serodiagnosis, a major problem in eradication campaigns. Because of this, rough Brucella mutants lacking the O-polysaccharide have been proposed as vaccines. Methodology/Principal Findings: To examine the possibilities of rough vaccines, we screened B. melitensis for lipopolysaccharide genes and obtained mutants representing all main rough phenotypes with regard to core oligosaccharide and O-polysaccharide synthesis and export. Using the mouse model, mutants were classified into four attenuation patterns according to their multiplication and persistence in spleens at different doses. In macrophages, mutants belonging to three of these attenuation patterns reached the Brucella characteristic intracellular niche and multiplied intracellularly, suggesting that they could be suitable vaccine candidates. Virulence patterns, intracellular behavior are lipopolysaccharide defects roughly correlated with the degree of protection afforded by the mutants upon intraperitoneal vaccination of mice. However, when vaccination was applied by the subcutaneous route, only two mutants matched the protection obtained with Rev 1 albeit at doses one thousand fold higher than this reference vaccine. These mutants, which were blocked in O-polysaccharide export and accumulated internal O-polysaccharides, stimulated weak anti-smooth lipopolysaccharide antibodies. Conclusions/Significance: The results demonstrate that no rough mutant is equal to Rev 1 in laboratory models and questions the notion that rough vaccines are suitable for the control of brucellosis in endemic areas.