The Pastorex((R)) (BioRad) rapid agglutination test is one of the main rapid diagnostic tests (RDTs) for meningococcal disease currently in use in the "meningitis belt". Earlier evaluations, performed after heating and centrifugation of cerebrospinal fluid (CSF) samples, under good laboratory conditions, showed high sensitivity and specificity. However, during an epidemic, the test may be used without prior sample preparation. Recently a new, easy-to-use dipstick RDT for meningococcal disease detection on CSF was developed by the Centre de Recherche Médicale et Sanitaire in Niger and the Pasteur Institute in France. We estimate diagnostic accuracy in the field during the 2006 outbreak of Neisseria meningitidis serogroup A in Maradi, Niger, for the dipstick RDT and Pastorex((R)) on unprepared CSF, (a) by comparing each test's sensitivity and specificity with previously reported values; and (b) by comparing results for each test on paired samples, using McNemar's test. We also (c) estimate diagnostic accuracy of the dipstick RDT on diluted whole blood. We tested unprepared CSF and diluted whole blood from 126 patients with suspected meningococcal disease presenting at four health posts. (a) Pastorex((R)) sensitivity (69%; 95%CI 57-79) was significantly lower than found previously for prepared CSF samples [87% (81-91); or 88% (85-91)], as was specificity [81% (95%CI 68-91) vs 93% (90-95); or 93% (87-96)]. Sensitivity of the dipstick RDT [89% (95%CI 80-95)] was similar to previously reported values for ideal laboratory conditions [89% (84-93) and 94% (90-96)]. Specificity, at 62% (95%CI 48-75), was significantly lower than found previously [94% (92-96) and 97% (94-99)]. (b) McNemar's test for the dipstick RDT vs Pastorex((R)) was statistically significant (p