You cannot know a computer has a virus by feel. You know by matching observable behavior and forensic artifacts against known malware patterns, then confirming with an on-demand scan, startup inventory, and network check. A single slow app proves nothing; clustered symptoms plus a confirmed detection do. Most "infected" machines are actually misconfigured, failing, or adware-laden. This guide separates the three, because each needs a different fix.

The detection mechanism: what malware actually does to your system

Malware is software, so it leaves software traces. Every process on Windows or macOS is registered with the operating system kernel, carries a file path, a parent process, and a signature status. A virus, trojan, or infostealer must persist across reboots to survive, which means it writes an entry somewhere the OS reads at startup. That entry is the artifact investigators hunt, because persistence is harder to hide than a running process.

Persistence lives in a finite set of locations. On Windows these include the Run and RunOnce registry keys, the Startup folder, scheduled tasks, services, and WMI subscriptions. On macOS they include LaunchAgents, LaunchDaemons, login items, and configuration profiles. Modern macOS also enforces notarization and Gatekeeper checks, so unsigned launch daemons are a genuine red flag. Security tools enumerate these locations and compare each entry against known-good databases and reputation services.

Detection engines use two complementary methods. Signature scanning matches file bytes against a database of known malware hashes and patterns, which is fast and precise but blind to new variants. Heuristic and behavioral engines watch what code does: encrypting many files quickly, injecting into another process, or opening a network connection to a known command-and-control server. Microsoft Defender, the built-in Windows engine, runs both continuously and quarantines on match. The practical implication: a detection alert from a reputable engine is evidence. A slow browser is not.

Quantified context: how common infections are, and what "slow" really means

Numbers keep this honest. The FBI's Internet Crime Complaint Center logged 859,532 complaints in 2023, with reported losses above USD 12.5 billion, and malware sits inside many of those cases rather than standing alone. Ransomware is the loudest subtype: an infection that encrypts files typically announces itself within minutes to hours by renaming or locking documents, so a silent weeks-long slowdown is rarely ransomware.

Performance expectations matter more than scare statistics. A healthy Windows 11 machine with 8 GB of RAM idles with roughly 150 to 250 processes and 40 to 60 percent memory pressure under light browsing. A CPU pegged at 100 percent for more than 10 minutes at idle is abnormal. Disk usage sustained above 80 percent on a system drive slows everything, and that cause is usually indexing, updates, or a failing drive, not malware. Browser slowdowns most often trace to extensions: an audit of extension permissions usually finds the culprit before any antivirus does.

Costs of cleaning up are concrete too. A consumer antivirus subscription runs about USD 30 to USD 100 per year. A professional malware removal service typically charges USD 100 to USD 300. Data recovery after ransomware, when possible at all, commonly exceeds USD 1,000. These figures explain why the practical-actions section below emphasizes fast, cheap verification before expensive intervention.

Symptom clusters versus single quirks

Isolated oddities are normal. One app crashing, a fan spinning up during video calls, or a browser tab freezing happens on clean machines weekly. The signal is clustering and timing: two or more independent categories appearing within days of each other, especially after installing software or opening an attachment. Use this comparison to separate the three usual suspects.

Signal Typical malware pattern Typical hardware or config cause Typical adware or PUP cause
CPU at 100% at idle Cryptominer, often with GPU spike Failing fan, thermal throttling Browser toolbar updater
New startup entries Random-named executables in AppData Driver or update service Bundled "helper" apps
Browser homepage changed Trojan proxy settings Sync account misconfiguration Hijacked extension
Antivirus disabled itself Active defense evasion License expired Fake security app
Network traffic at idle Beaconing to unknown IPs Cloud sync, backups Telemetry from toolbars
Files renamed or encrypted Ransomware Disk corruption None

Read the table as a probability guide, not a verdict. The counterintuitive detail: adware and "potentially unwanted programs" outnumber true viruses on consumer machines by a wide margin, and they produce cleaner, more consistent symptoms than malware does. Malware tries to hide; adware wants your attention. If pop-ups are the main complaint, you are probably dealing with a PUP, and the fix is uninstalling programs and resetting the browser, not reinstalling Windows.

Timeline: what a real infection looks like hour by hour

Malware follows a lifecycle, and each stage leaves different evidence. Knowing the clock helps you decide whether a symptom today connects to an event last week.

Time since infection Stage Observable artifact
0 to 10 minutes Delivery and execution New file in Downloads or Temp; browser warning bypassed
Minutes to 1 hour Persistence established New startup entry, scheduled task, or launch agent
1 to 24 hours Payload activity Beaconing traffic, credential access, or encryption begins
1 to 3 days Defense evasion Antivirus exclusions added, security tool disabled
3 to 14 days Lateral or data movement Unusual logins, large outbound transfers
Weeks Dormancy or monetization Silent, low-bandwidth traffic; occasional CPU spikes

The professional caveat sits at the bottom of that table: sophisticated malware, especially on managed corporate devices, can stay dormant for weeks and leave almost no user-visible symptom. Absence of symptoms is not proof of cleanliness, which is why scheduled scans and monitoring matter more than periodic suspicion.

Red flags that justify immediate action

Most warnings are noise, but a few deserve an immediate response before any diagnosis.

Ransomware note on screen or files renamed with unfamiliar extensions: disconnect from the network now, do not pay, and preserve the machine for recovery. Your antivirus or built-in security tool has been disabled and you did not disable it. A browser warns that your connection is not private on sites that worked yesterday, paired with a changed proxy or DNS setting. Unexpected password reset emails, new login alerts, or bank notifications: assume credential theft and change passwords from a different device. A new remote-access tool you did not install, such as a screen-sharing app, especially after a phone call pressuring you to install it. Sustained outbound traffic at idle to an unfamiliar IP address or country, confirmed in a network monitor.

The phone-call scenario deserves emphasis. Tech-support scams that convince you to install remote-access software are one of the most common consumer incidents, and the "virus" is often the remote tool itself. Legitimate companies do not call about your computer.

Practical actions: a step-by-step verification protocol

Run these in order. Stop when you have a confirmed detection or a clean bill of health from two independent methods.

First, check built-in protection status. On Windows, open Windows Security and confirm real-time protection is on and the last scan is recent. On macOS, confirm Gatekeeper and XProtect are active and check System Settings for unknown configuration profiles. A disabled engine is itself a finding.

Second, run a full on-demand scan with the built-in engine. Microsoft Defender's full scan takes 30 to 90 minutes on a typical machine. Follow with a second-opinion scanner from a reputable vendor, since engines disagree on borderline files. Two clean full scans from independent engines is meaningful evidence; one is weaker.

Third, inventory startup items. On Windows, review Settings, Apps, Startup and Task Manager's Startup tab, then check scheduled tasks and services for entries with random names or paths in user folders. On macOS, inspect Login Items and the LaunchAgents and LaunchDaemons folders. Compare each entry against software you recognize. Unknown does not mean malicious, but it means investigate.

Fourth, audit browsers and extensions. Remove extensions you did not deliberately install, reset the homepage and search engine, and check proxy settings. This step resolves the majority of pop-up and redirect complaints.

Fifth, check network behavior. Look at active connections and note any process talking to an unfamiliar address at idle. Consumer network monitors show this in plain language; a corporate device should be handed to IT instead.

Sixth, decide the threshold for escalation. If two independent scans are clean, startup is understood, and network traffic is quiet, treat the machine as clean and investigate hardware or configuration for remaining slowness. If any scan detects malware, let the engine quarantine it, then rescan; if detection recurs after quarantine, back up personal files and reinstall the operating system from official media. Reinstallation is the only method that guarantees removal.

Finally, change passwords for email and financial accounts from a different, known-clean device if you suspect credential exposure. Enable multi-factor authentication everywhere it is offered.

Limitations: what scans cannot prove, and when to get help

No scanner detects everything. Signature engines miss new malware until signatures ship, often days after a variant appears. Behavioral engines produce false positives and false negatives alike. Rootkits and firmware-level implants can survive operating system reinstallation, though these are rare outside targeted attacks. A clean scan is strong evidence, not proof.

Encrypted traffic limits network inspection on consumer machines, and cloud-delivered malware may never write a persistent file. Managed corporate devices carry monitoring the user cannot see, so a "clean" personal scan says little about a work laptop; route those to IT. If you are unsure after the protocol above, a professional diagnostic typically costs USD 100 to USD 300 and includes deeper forensic checks than consumer tools provide.

The honest bottom line: cluster your symptoms, verify with two independent engines, inventory persistence, and escalate on confirmed detection. Everything else is suspicion, and suspicion alone should not cost you a reinstall.

This article is general educational information about malware detection, not a diagnosis, treatment plan, or security guarantee. It cannot confirm whether a specific device is infected, and no scan or procedure described here promises complete removal or protection. For a device holding sensitive data, or after suspected credential theft, consult a qualified local cybersecurity professional or your organization's IT security team.