I'm going to assume you are trying to reduce the chance that your software will fall to an attack. It's a big problem.
From most effective to least effective then,
-- Don't use system software with holes. Your operating system, for instance. Windows (and MS-DOS before it) has a truely miserable security record. The Unix emulators (Linux, xxxBSD, most prominent among them, but Mac OS X is based on BSD) are not only free (ie, no charge, but also free as in you can do what you like -- no shrinkwrap 8 page license contracts; except Mac OS X and even that has an open source counterpart, Darwin) and come with hundreds of software application packages right out of the box, but have much better security records. OpenBSD is most likely the best as one of its chief goals is security and lack of holes.
In all cases, Unix to Windows and on through to everything else, there is a requirement that someone manage the installation -- install software, check that default protection choices are correct, ensure that passwords are sensible ('password' is an outstandingly stupid choice of password, so is 'rover,' 'fido', 'fluffy', your own user name, your auto license plate nr, your phone nr (any of them), your significant other's name, ...) and so on. It's just that Windows claims to be user friendly and trouble free (not true for any version so far) and makes administration obscure and awkward at best. Linux and the BSDs make no such claim and are much more honest in that respect. Someone trustworthy will have to pay attention to the security side of things -- what has just been discovered this week about holes in some existing piece of software, what patches have been released for the problems found last week, ... And this for all the software installed on your machien. This will take time and effort and understanding of your equipment as a computer with software, none of which are most people really happy with. There's little time, and not much available effort, and I don't really want to understand this stuff on its own terms, (eg, "I want to use the word processing software as a really capable typewriter. I'm an avant garde poet, not a computer science geek). The answer to this is just the same as Plato is said to have given to some royal son when geometry was the rocket science of its day. There is no royal road to geometry, or in modern terms, sorry Charley, there's no low effort / high security / low knowledge approach to security.
-- Disconnect your systems from the Internet, for there are assorted malicious cretins around the world who would like to infect your machine (for various reasons) or take it over (ie, make it a zombie) to serve as a spam or porn distribution node. If there's an air gap between your machine and the Internet (and no wireless connection either), these morons will be blocked.
-- Short of that, if you must be connected to the Internet, get, install, and correctly configure quality firewall software on your machine or on a bastion machine (ie, one between your machine and the Internet which runs nothing but a firewall). Linux and the BSDs come with excellent firewall software builtin, you need only configure it properly. Which requires only a good understanding of Internet protocols and attack possibilities, combined with an ability to assemble machines with lots of small parts.
A proxy server (or several) for each of the protocols you will be using will block a class of attacks no firewall can (slipping malware past a firewall hidden within a permitted protocol's data), and are generally useful as well.
Remove or disable vulnerable utility software (eg, ftp has a terrible reputation for insecurity and the 'r' utilities cannot be made secure, use SSH instead). And do the same for applications (word processing or databases or image editing software). Run only those services (ie, daemons, or servers, or some other name for this sort of thing) on your machine that you actually need. Windows has started the Windows Messenger Service by default in every installation -- it's not only insecure to date, but isn't used by much of anything. Every piece of permanently running software on your machine is an invitation to an attacker, if there is an exploitable hole in that software. Especially if the software is of the kind that listens for Internet traffic it can provide a service for.
Since bugs, and new attack schemes, are regularly uncovered (daily for Windows, weekly/monthly for Linux/BSD systems), it is a very good idea to install patches (ie, fixes for the bug or newly discovered vulnerability) as they become available This suggests that you should use an operating system only from a vendor or group that actively keeps it current with released patches. Microsoft releases patches for its products for a few years after they're released, but like OS administration, they're often not very transparent. Microsoft likes to include new features, new products, and new license terms in its various Servic Pak releases. The Linux / BSD world is much better (the patches are released with documentation in most cases, so you can figure out just what is going on, but also are generally available in source code which prevents, if you look, ignorance about the function of the patch.
Browsers have been a particular problem. The Web is so complex (multiple protocols, for multiple purposes, multiple standards for software from multiple sources, ...) that security is difficult. Nevertheless, Internet Explorer has earned high marks for insecurity. It has been non-standard, and riven with security problems throughout its lifetime. It has been so bad that various US government departments have officially suggested it not be used at all. Microsoft has embedded parts of the operating system in the Internet Explorer libraries and so it can't really be removed. All one can do is to not use it as a browser. More of less the same is true of Outlook and Outlook Express, Microsoft's email management programs. They have been security nightmares a well. Alternatives exist which are considerable improvements. For email, Thunderbird is free on all platforms, Eudora is free in a limited functionality version, and so on. For browsing, Opera is free in a limited functionality version, Firefox is free in all versions, and so on. None is a problemtic as Microsoft's offerings have been.
-- But assuming that your attempts ot be careful have failed, you will need to detect and remove malware that has managed to get through even so. There are free anti-virus programs, some of which can scan all incoming content for problems. This function has been fraught with incompatibilities on Windows in the past, so it may be more trouble than it's worth. AVG from Grisoft, and Avast from Alwill have good reputations amongst the free (limited functionality) offerings. The famous commercial offerings have varying reputations.
But there is an additional problem with malware. Let's say you sign up for something (like Kazaa). Part of the license may be that you allow some spyware (which gathers data for marketing purposes) on your machine. Should this, if detected, be treated as a virus and deleted? If so, Kazaa (and others) claim they will stop working. Spyware removers, unlike most of the anti-virus vendors, take the position that this stuff is malware and attempt ot detect and remove it. Regardless of what happens with Kazaa (or whatever). There are some well regarded free spyware removers for Windows; Spybot.Search and Destroy is one, Lavasoft's Adaware is another.
Note that there is no market for anti-virus sfotware on Linus and the BSDs. There have only been a very few ever seen. The anti-virus products for Linux and the BSDs are primarily for detecting an removing Windows viruses, since many Linux or BSD installations act as central file stores for a herd of Windows machines. Likewise there isn't much of a spyware problem for Linux or the BSDs, so there isn't much, if any, spyware removal software available for them.
-- There's a philosophical problem as well with your question. Software systems are the most complex things human beings have ever built. We are operating on (or mostly beyond) the edge of our competence just getting things to work. Making certain they are proof against attack is inherently impossible. New attacks, previously unforeseen, can't be anticipated. So making your software proof against cracks can't be like sprinkling security dust over it. Even if there were such stuff, no one (not me, not you, not the bulging brains at Bell Labs) knows which kind to sprinkle just now. It'll probably be different next week, next month, next year. All an alert security manager (for most systems that means the owner / operator) can do is to take thaose steps which increase security, if not guarantee it. I've outlined a goodly number here. Paying attention to the the literature, the news groups, the news feeds, and so on, and taking sensible action as a result is the next step, one that will never be, and can never be, completed.