The Short Version
An oxidation number is a bookkeeping tool that tracks how many electrons an atom effectively gains or loses when it bonds with other atoms. It tells you whether something was oxidized or reduced in a reaction, and it's the reason we can balance redox equations without having to track every single electron by hand. The actual calculation is straightforward if you know the default rules. Hydrogen is +1 when bonded to nonmetals, -1 when bonded to metals. Oxygen is almost always -2, except in peroxides where it's -1. Fluorine is always -1. Group 1 metals are +1, Group 2 are +2. For anything else, you assign electrons based on electronegativity and work backwards from the overall charge of the molecule or ion. Take dichromate, Cr2O7 2-. Oxygen is -2 each, so that's 7 times -2, which is -14. The whole ion is -2, so the two chromium atoms together have to be +12. Each Cr is +6. That's it. The math doesn't get more complicated than that in 90 percent of undergraduate problems.
Where people mess up is assigning the rules correctly under pressure during exams or lab reports. The most common error I see is students assigning oxygen as -2 in a peroxide without checking the structure first. Hydrogen peroxide is H2O2, and if you blindly apply the oxygen rule you get hydrogen at +1 and oxygen at -2, which sums to -2 for the whole molecule. The molecule is neutral though, so you immediately know something is wrong. The fix is to check whether the O-O bond exists, which means it's a peroxide and oxygen is -1 instead.
Why We Use It Instead of Something Better
Oxidation numbers are formal charges, not real physical quantities. The atom in a covalent bond doesn't actually lose or gain a full electron. It's a hypothetical construct that works because it maps cleanly onto electron transfer patterns. When you're balancing a reaction between permanganate and iron(II) in acidic solution, you don't need quantum mechanics to figure out that Mn goes from +7 to +2 and Fe goes from +2 to +3. The oxidation number gives you that information in two seconds. The alternative is tracking actual electron density through molecular orbital theory, which takes considerably more time and rarely gives you a cleaner answer for routine stoichiometry work. For practical purposes, oxidation numbers are the right tool. They are not the right tool when you need to predict magnetic properties, color spectra, or reaction kinetics. Don't try to stretch the concept beyond what it was designed for.
Get the Full Details
The Rules You Actually Need to Memorize
Elemental form is always zero. O2, N2, Fe(s), S8. If it's sitting alone on the periodic table's left side of a reaction arrow, its oxidation number is zero. This catches people out because they see oxygen and immediately think -2. Monatomic ions carry their charge. Na+ is +1. Ca2+ is +2. Cl- is -1. This one is trivial but easy to overlook when the ion is buried inside a larger compound. Halogens are -1 unless bonded to oxygen or a more electronegative halogen. In NaCl, chlorine is -1. In ClO3-, chlorine is +5 because oxygen takes priority. In ClF, fluorine is -1 and chlorine is +1 because fluorine is more electronegative than chlorine.
The sum equals the overall charge. This is the constraint that lets you solve for unknowns. A neutral molecule sums to zero. An ion sums to its charge. That's the entire balancing act.
A Problem That Isn't Obvious
I spent about forty minutes once trying to balance a reaction involving thiosulfate, S2O3 2-, against iodine. The issue was figuring out the oxidation state of sulfur in thiosulfate. The two sulfur atoms are not equivalent. One is bonded to three oxygens and the other sulfur, and the other is bonded only to the first sulfur and carries a terminal charge distribution that makes a simple arithmetic approach give you an average oxidation state of +2 for both sulfurs. That average is chemically meaningless for balancing purposes. The workaround was to look at the structure. The central sulfur (the one bonded to three oxygens) is in a sulfate-like environment, so it's +5. The terminal sulfur, which is bonded only to the central sulfur, is -1. When thiosulfate reacts with iodine, only the terminal sulfur gets oxidized to elemental sulfur, while the central sulfur stays at +5. If you use the average of +2 for both, you balance the equation incorrectly and your stoichiometry is off by a factor that matters in titration work. X-ray photoelectron spectroscopy data confirms the two distinct sulfur environments, but for most purposes, drawing the Lewis structure and assigning by position is fast enough.
When Oxidation Numbers Fail Completely
Bonding in transition metal complexes is where this system starts breaking down. Consider [Fe(CN)6]4-. If you assign carbon as -4 and nitrogen as +3 based on electronegativity arguments through the ligand, you get iron at +2, which happens to be correct. But in [Fe(CO)5], carbon monoxide is a neutral ligand, so iron is formally zero. Is iron really in the zero oxidation state, or is there significant back-bonding that makes the formalism misleading? The answer is that the oxidation number is still +0 by convention, but the actual electron distribution is more complex. The convention holds because it gives consistent bookkeeping across similar reactions, not because it describes physical reality accurately. Organometallic chemistry has a whole set of workarounds for this. The covalent bond method, the ionic method, and the donor-pair method will sometimes give different oxidation states for the same compound. None of them are wrong. They're just different conventions for different purposes. Pick the one your lab uses and stick with it. Another case where oxidation numbers are useless is in solid-state materials like spinels and perovskites. Mixed-valence compounds such as magnetite, Fe3O4, contain iron in both +2 and +3 states. The average oxidation state is +8/3, which is not a number any chemist would ever write down in a report. You have to know the crystal structure to assign the individual sites correctly.
How to Actually Use This in a Reaction
Identify which atoms change oxidation number between reactants and products. Assign the numbers on both sides. The atom whose number increases is oxidized. The atom whose number decreases is reduced. Multiply the half-reactions so the electron count matches. Add them back together. Check that atoms and charge balance. This takes about three minutes for a standard undergraduate problem and about twenty minutes if you're dealing with something like the reduction of manganate in basic solution where multiple products are possible. The real value shows up when you're writing mechanisms or predicting whether a reaction is thermodynamically feasible. A large change in oxidation number usually corresponds to a large driving force, though that correlation breaks down when kinetics are slow or when the reaction pathway involves high-energy intermediates. Don't confuse thermodynamic favorability with reaction speed. If you need to look up oxidation states for unusual compounds, the CRC Handbook of Chemistry and Physics and the NIST WebBook have tables that cover most inorganic species. For organometallics, the Housecroft and Sharpe textbook is reliable, though it assumes you already understand the conventions it's using. There is no single authoritative source that covers everything because the conventions themselves are not always consistent across subfields.
