Referee 1's review (JF Schoeman):

The main objective of this study by Torok et al was to "determine the clinical and microbiological features, outcome and baseline variables predictive of death, in HIV-associated TBM".

A strength of the study is that it was conducted prospectively in a relatively large cohort of HIV-associated TBM cases. The very high rate of CSF smear and culture positivity is recommendable. We agree with the authors that the large volumes of CSF and the dedicated technician definitely played a role. The patients in this study, however, were severely immune-compromised. Was this comparable to the immune status of the patients in other studies quoted by the authors where the culture rates were much lower?

The main shortcoming of this study is that it does not increase our understanding of the pathogenesis of the poor outcome in HIV-associated TBM. The authors state that early deaths were mainly as a result of TBM and that the cause of late deaths was often not known. Most of the early deaths occurred while the patients were under direct observation. Yet no information is given on the incidence and treatment of hydrocephalus and the incidence of new paralyses indicative of new cerebral infarcts. What role did extra-cranial factors (e.g. nutritional state, pulmonary TB etc) play in the high mortality rate? Why did so many (50%, Table 1) of stage 1 TBM cases die? The extremely high mortality rate in this study (even in the cases with stage 1 and 2 TBM) needs some clarification.

A few minor points:

• The category "possible TBM" last paragraph page 7 should be omitted since it is not referred to again in the text and is not mentioned in Table 1.
• Reference 25 (for the staging of TBM of TBM) refers to the original MRC classification and not the "modified' classification.
• The relationship between death and CSF polymorph count and serum sodium should be interpreted in the context of the stage of TBM in these patients.
• The last sentence last paragraph page 12: "appeared to be more common". Rather "was more common although the difference was not statistically significant". The relationship between drug resistance patterns and outcome found in this study needs to be briefly discussed. Are the findings in line with the literature?
• Table 1 can be shortened and needs some clarification. The resistance patterns and their relationship to death can be interpreted and summarized in the text in a few sentences. The authors should rather refer to the latest terminology regarding drug resistance as suggested by the WHO in 2006. E.g. mono DR, poly DR, multi DR. The numbers in some of the rows do not add up (possibly because of patients that were lost to follow-up?).

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.