Math.max and Math.min NaN on undefined entry

I assume the undefined is actually some variable.

You can substitute -Infinity for any NaN value to ensure a number.

var foo;

Math.max(5, 10, isNaN(foo) ? -Infinity : foo); // returns 10

Same concept can be used on Math.min, but with Infinity:

var foo;

Math.min(5, 10, isNaN(foo) ? Infinity : foo); // returns 5

Write your own max function:

function max() {
    var par = []
    for (var i = 0; i < arguments.length; i++) {
        if (!isNaN(arguments[i])) {
            par.push(arguments[i]);
        }
    }
    return Math.max.apply(Math, par);
}

Or shorter using array filtering:

function max() {
    var args = Array.prototype.slice.call(arguments);
    return Math.max.apply(Math, args.filter(function(val) {
       return !isNaN(val);
    }));
}

Usage:

max(5, 10, undefined);   // output: 10

Check that each number is defined before passing it to max. Use an sentinel value as a default (such as -Infinity). For min you would want to use Infinity:

var a = 5;
var b = 10;
var c = undefined;

var defaultForMax = function (num) {
    return typeof num === 'number' ? num : -Infinity;
};

c = defaultForMax(c); // c is now -Infinity

Math.max(a, b, c); // 10

http://jsfiddle.net/QTckE/

It's as simple as:

// ES6 (returns: 10)
Math.max(...[5, 10, undefined].filter(Number.isFinite))

Or:

// ES5 (returns: 10)
Math.max.apply(null, [5, 10, undefined].filter(Number.isFinite))

I think it is readable, and simply conveys its purpose.

Explanation: We take advantage of the Number primitive wrapper object, to filter out the undefined values, and then pass it to the Math.max function, as arguments with the spread operator in ES6, or with the Function.prototype.apply() method in ES5.