Its a problem inherent to the dye. Have been experiencing that the extent of precipitation varies from run to run which varies the concentration of neutral red in the solution. As a result, i am finding huge variations in od values from the same number of cells. Have been working on ways to minimize precipitation. As soon as i manage something, will let you know.
Do you incubate your neutral red dye solution in culture medium at 37oC prior to adding to assay plates? What i do is i prepare a 500microg/ml solution of neutral red in RPMI containing inactivated serum at 10%. Of course, my exposures are in medium containing 1% serum. I incubate the above neutral red solution at 37oC overnight. There will be a little precipitate, but not much. I spin it at 4000rpm for some 5mins, and take the clear supernatant. My cell exposure is in 200microL, when its time for neutral red assay, I add 20microL of the above neutral red soln (after centrifigution) and incubate the plate for an additional 3h. This is followed by discarding of medium, washing the cell layer a couple of times with PBS (as gently as possible) and then extracting the dye in a soln contining ethanol:h20:acetic acid in 50:49:1 ratio. This procedure has been helpful.
It comes down to pH, slightly higher the neutral will cause the ammonium (-NH4+) moiety to deprotonate (-NH3) thus making the molecule non-polar i.e. not water soluble and crystallizes out of solution.