You said above "That, and we observe mutations in every generation"
I'm looking for a ballpark figure..
What percentage of mutations are considered to enhance the fitness of an organism?
It varies from species to species. There is no set percentage of mutations that are beneficial, and the same mutation that is beneficial in one context may be harmful in another, such as the ability to breathe salt water. Useful if the creature lives near and spends a lot of time in salt water, but completely useless or even deadly to one that lives in a desert. Some mutations are a little bit of both, such as the one for sickle cell anemia. It significantly increases resistance to malaria in both homozygotes (people afflicted with the condition) and heterozygotes (rarely affected by the mutation so much as to suffer from it, but still have the benefits of the malaria resistance), but it also produces a potentially deadly condition in people with two copies of the mutated gene. It just so happens that malaria resistance is so beneficial, that the negatives were not a strong enough selection pressure to have it decrease in frequency in the population.
If you want a "ballpark figure" for this stuff, sorry to say, you won't find one. The variation in it is too extensive. What I can tell you is the number of mutations in your DNA that neither of your parents have: between 40-60. Using the estimated number of births per day in 2014, and using the average of 50 independent mutations per person, that's the potential for a whopping 17,650,000 new mutations in the human population PER DAY. Even if only .00001% of those were on genes, that would make a new non-neutral gene occur each day. And humans have a pretty long reproductive cycle, and nothing is special about our mutation rate. That number would be exponentially larger for bacteria, which reproduce far faster and have higher mutation rates than humans. And I am being the least generous I can be without being a parody of criticism. Neutral mutations may dominate, but not to that extent.
BTW, that 70% comes from the pool of mutations affecting actual genes, so it doesn't include mutations on noncoding regions. Yes, many mutations even on coding regions of DNA are neutral.
What percentage of mutations would be considered as deleterious?
Just so you know, deletions are not inherently bad mutations. In fact, I know of at least 2 which were necessary for humans to become as intelligent as they are; one for developing a jaw muscle that would have restricted cranial capacity but made our jaws stronger, and another that regulates brain growth. Both were mixed bags; on the good side, we became that much smarter, and intelligence is highly beneficial to survival. On the bad side, our jaws are weaker than other primates, and we are more prone to brain cancer. Most mutations that affect phenotype are like that.
As for estimates given on an actual percentage, I couldn't find it for humans, but in fruit flies, the percentage of mutations that ended up having a measurably negative impact was 70%. That means 30% are beneficial or neutral. To be blunt, it is neither shocking, nor a problem for evolution, that the majority of mutations aren't beneficial. However, it should be noted that fruit flies don't have as many protective mechanisms against genetic damage as humans do, namely, the degree of redundancy in their genome is less than is seen in humans, and their genome is smaller, making a mutation more likely to occur on a coding portion. As I have mentioned before, how mutations affect species varies from species to species thanks to so many factors that it is impossible to say what percentage of mutations in humans are harmful by using the percentage seen in fruit flies.
What percentage would be considered as neutral?
The majority are, but even if the number of beneficial mutations was abysmally low, less than .000000001%, natural selection and evolution would still apply.