Very different answer: your secondary and/or primary bound unspecifically and over time they diffuse off of the weak binding site? Since you take care of bleaching with all the right conditions and measures, why would the staining disappear? Is it just black now or do you have overall diffuse background? How is the tissue mounted? 

Are you protecting you slides (I'm assuming you're using slides or a multi-well plate) from light after you're done treating? Fluorescence antibodies are light sensitive and will fade rather quickly when exposed to ambient light. Even in protected sections, fluorescence will fade over time due to degradation of the signal. I would recommend placing your slides in a slidebox and wrapping the box in aluminum foil to reduce the chances of photobleaching from occurring.

thanks. although i did not mentioned it, we take the right precaution in order to avoid this.

the sections are been mounted on slides with glycerol:PBS as an anti fading agent. 

any other ideas???

Try commercial anti-fading mounting solutions from molecular probes . They work great

All the above suggests are good. We use an anti-fading media that consists of 100 mg p-phenylenediamine in PBS/glycerol that is second to none in preventing fading. Keep it in the freezer (-10 C) and it keeps forever. See J. Immun. Method. 43:349-350, 1981.

Very different answer: your secondary and/or primary bound unspecifically and over time they diffuse off of the weak binding site? Since you take care of bleaching with all the right conditions and measures, why would the staining disappear? Is it just black now or do you have overall diffuse background? How is the tissue mounted? 

Heat and light are the main culprits for fading.  How are you storing your sections after staining?  -4C? -20C? Left at room temperature they can fade pretty fast. 

I agree with Karen about heat and light as your main concerns but would add a third and that is time.  As you know, antigen antibody reactions are not mediated by covalent bonds rather they are "weak interactions".  As a result, antigen-antibody complexes will dissociate over time.  This process can be slowed by storing samples in the refrigerator or freezer but it cannot be stopped.  A major factor that will determine the dissociation rate is the affinity of the antibodies for the antigen.  The higher the affinity, the lower the rate of dissociation.  In general, polyclonal antibodies have higher affinities than monoclonal antibodies and IgGs have higher affinities than IgMs.  So if you are using a single step procedure with fluorochrome bound to your primary antibody and your antibody is an IgG you will have the least amount of dissociation over time.  If you're using a two-step or three-step procedure where the final antibody is bound to fluorochrome and any of the antibodies is an IgM, then the dissociation rate will be much higher.  Finally, you should also consider the pH and ionic strength of your mounting solution as these can also have important effects on the dissociation rate.

Might be something simple enough as not enough rinsing after the last step before cover slipping.  The pH of the tissue might be out of range slightly at the end and make the fluorescence fade over time.   

There is more.  1) If you are using a direct tagged secondary to get fluorescence, you are using the LEAST sensitive of all fluorescent techniques, and then having very little product means it will go away more rapidly.  see  paper ATTACHED.  In addition, which fluorophore you use determines in part, how easily it will fade.  In my experience some reds (Texas Red), other Alexa Dyes (Alexa orange), or Cy3  fade less when compared to fluoroscein, or Rhodamine.  CY2 fades a bit more than the orange or red; Oregon Green was better.  When we prepare slides for fluorescence viewing we dry sections on the slide, then quickly scale through alcohols, dehydrate well and place in xylenes, and then a low fluorescent mounting medium and the fluorophores do not fade.  For water-soluble mounting media, you should avoid just glycerol and use those products with anti-fade (reducing) agents (even n-propylgamine worked better than just glycerol.  That said, fading will be far less of a problem if the amount of fluorophore is maximized.  Using TSA amplified fluorescence as your standard not only saves primary antibodies (and all other relevant reagents) but gives up to 100x the product for each molecule of primary antibody bound (see figure on page 2.12.7. The slide in that micrograph was acquired in 2007 and still looks the same for the TSA amplified approach. With zylene based media, the key is being sure that the sections are water free.  With water in them, they will fade.