Progressive multifocal leukoencephalopathy (PML), a demyelinating disease of the central nervous system caused by the human polyomavirus JC (JCV), results from a lytic infection of the oligodendrocytes that typically occurs in profound immunosuppression. Characteristically, the inflammatory response is not observed on histopathological findings, and neuroimaging studies rarely show contrast-enhancing lesions [1]. Atypical presentations of some diseases arising from pre-existing infections and inflammatory reactions against several opportunistic pathogens have been described in AIDS patients soon after starting HAART [2]. We describe a case of PML with an inflammatory response occurring shortly after initiating HAART that was misinterpreted as neurotoxoplasmosis. A 32-year-old Brazilian man naive of antiretroviral treatment was diagnosed HIV-1 positive in May 2004 as a result of chronic diarrhoea and weight loss. The physical examination disclosed oral candidiasis and seborreic dermatitis and his CD4 cell count was 18 × 106 cells/μl. He promptly received a combination of zidovudine, lamivudine, and efavirenz, as well as prophylaxis against Pneumocystis carinii and Toxoplasma gondii. Three weeks later, he was admitted because of mental confusion and left hemiparesis. His CD4 cell count was 146 × 106 cells/μl and his viral load was 750 000 copies (5.87 log/ml). A brain magnetic resonance imaging (MRI) scan revealed a large white matter lesion on the right parietal lobe without gadolinium enhancement (Fig. 1a and b). Empiric anti-Toxoplasma treatment was initiated. His clinical condition continued to deteriorate and 2 weeks later he was submitted to another brain MRI scan that showed an increase in the previous lesion and an irregular gadolinium enhancement (Fig. 1c and d). A lumbar puncture was performed and the analysis of the cerebrospinal fluid (CSF) disclosed 3 mononuclear cells/ml3, 65 mg/dl of proteins, 42 mg/dl of glucose, the presence of HIV-1 antibodies, and an IgG index of 1.05. JCV DNA was detected by a nested polymerase chain reaction (PCR) developed for the amplification of a 386 bp fragment of the T antigen region of the virus. Treatment for neurotoxoplasmosis was interrupted but the antiretroviral regimen was maintained. His neurological condition began to improve 3 weeks later, and he was discharged with a CD4 cell count of 261 cells/μl and undetectable viral load. A further brain MRI scan, performed 5 months later, showed the persistence of the white matter lesion but without gadolinium enhancement (images not shown) and a further CD4 cell count of 407 cells/μl.Fig. 1: Brain magnetic resonance imaging scans showing a large demyelinating lesion (a) that enlarged after 5 weeks of HAART (c) (flair images); at this moment gadolinium enhancement is observed (arrow in d; compare with b) (T1 with contrast).This report raises important observations that must be considered in the management of any AIDS patients presenting with unexpected neurological symptoms soon after initiating HAART. In our case, neurological symptoms appeared a few weeks after the introduction of HAART in a patient with profound immunosuppression. Even in the absence of the typical brain lesions of neurotoxoplasmosis, this patient was subjected to anti-Toxoplasma therapy. Although cerebral toxoplasmosis is a very frequent neurological complication in HIV patients, mainly in developing countries, it is more common in patients with milder immunodeficiency. Therefore, in cases of severe immunosuppression, other opportunistic infections and tumours should be appropriately ruled out [3]. In this setting and in the absence of a mass effect a CSF analysis with PCR could be a non-invasive and reliable method of obtaining the correct diagnosis. For example, with regard to PML, the PCR for JCV has a sensitivity ranging from 61 to 100% and a specificity of 99%. Although the incidence of opportunistic infections and tumours has decreased since the introduction of HAART, the incidence of PML remains stable [4,5]. PML cases appearing soon after starting HAART have been described and associated with immune reconstitution inflammatory syndrome [6–9]. Prominent perivascular inflammatory infiltration has been seen [6] and maybe this inflammatory reaction accounts for the unusual contrast enhancement observed [8], as seen in our case (Fig. 1d). Therefore inflammatory PML could be a characteristic of patients experiencing immune reconstitution of their CD4 cell counts during the first weeks of HAART. An interesting point is the relative good prognosis described in PML associated with immune reconstitution inflammatory syndrome, as in our case, and corticosteroids should be used only in severe inflammatory reactions with significant oedema and mass effect [10,11]. In conclusion, indiscriminate empiric anti-Toxoplasma therapy may no longer be appropriate as a first-line approach to all HIV-positive patients with focal brain lesions, even in developing countries. New diagnostic strategies should be considered to identify the underlying disorders, especially in atypical settings. In cases in which lumbar puncture is not contraindicated, the routine clinical usage of CSF PCR in AIDS patients should be encouraged to reach a correct diagnosis.