TGA Classification Rules: The Framework That Decides Your Regulatory Pathway

Every medical device entering the Australian market gets sorted into a risk class before anything else happens. That class determines which conformity assessment route applies, how much clinical evidence is required, and how long your ARTG inclusion process will take. This is important to know because if you get the classification wrong at the start, the rest of your submission timeline shifts with it.

What Does TGA's Classification System Actually Cover?

TGA sorts devices into four broad tiers: Class I, Class IIa, Class IIb, and Class III, running from lowest to highest risk. Where your device lands depends on a combination of factors rather than a single attribute. Intended purpose, duration of use, degree of invasiveness, whether the device is active or non-invasive, and which part of the body it affects all feed into the eventual outcome.

Duration of use serves as the foundation because it shows up across multiple rules. TGA defines transient use as continuous use under 60 minutes, short-term use as 60 minutes to 30 days, and long-term use as anything beyond 30 days. Temporary interruptions for cleaning, eating, or sleeping do not break continuity for these purposes.

Manufacturers Sometimes Get This Wrong

Most classification errors do not come from misreading a single rule. They come from applying the rules in isolation rather than as a set.

A device used in combination with another medical device gets classified separately from that device. An accessory is classified separately from the device it supports. And if software drives or influences a medical device, that software inherits the classification of the device it controls.

Why does this matter in practice? Because a single product can trigger several rules at once, and the regulations are explicit about what happens then: the highest resulting classification applies. The one exception is when devices are classified as export only Class I regardless of what other rules might suggest.

The Rules That Carry the Most Weight

Rule 1 applies to every medical device and establishes the duration-of-use baseline described above. Rule 2 acts as a catch-all, applying when none of the more specific rules fit, and typically covers non-invasive, non-active devices such as those collecting body fluids where return flow is unlikely.

Active devices, including software-based medical devices, sit under their own set of provisions in Rule 4. A device that supplies energy absorbed by the patient's body, excluding visible light, lands in Class IIa under Rule 4.3(2)(a). Devices that image the in vivo distribution of radiopharmaceuticals fall under the same class and rule subsection. Devices used for direct diagnosis or monitoring of vital physiological processes, where the risk is not immediate, also sit at Class IIa.

The classification escalates to Class IIb under Rule 4.3(3) when the stakes rise. Devices monitoring vital physiological parameters where variation could create immediate danger to the patient fall here, as do devices emitting ionising radiation for diagnostic or therapeutic interventional radiology, and devices that control or directly influence the performance of that radiation equipment.

Rule 4.5 deals with devices intended to diagnose or monitor diseases or conditions and ties the classification to public health risk. If the condition could lead to death or severe deterioration without urgent treatment, or poses a high public health risk, the device sits at Class IIb. A serious condition with moderate public health risk brings it down to Class IIa. Anything outside those categories defaults to Class I.

Tissue, cell, and substance-derived devices have their own pathway under Rule 5.5, and this is an area that has shifted recently. Devices containing materials of animal, microbial, or recombinant origin were historically pushed straight to Class III. Changes effective from 1 July 2024 removed the blanket reference to "specified medical devices" that previously locked these products into the highest risk tier, meaning some devices that were Class III may now classify lower depending on their specific composition and intended use. Sponsors with existing Class III ARTG entries affected by this change have transition timelines running through to 1 July 2026, after which supply requires the device to be correctly classified under the amended rule.

Software Classification Is Its Own Conversation

The TGA published dedicated guidance on classifying active medical devices, including software-based medical devices, with the first version released in October 2021. The guidance is explicit that it does not cover every classification rule and is not a substitute for working through Schedule 2 in full. What it does provide is clarification on how the core rules apply when the device in question is software, particularly where that software drives or influences the function of other medical devices.

The responsibility for getting this right sits with the manufacturer. TGA's role is not to classify your device for you; it is to assess whether the classification you have assigned, and the conformity assessment evidence that follows from it, meets the regulatory bar.

How Classification Connects to Everything Downstream

Classification is not a box-ticking exercise that happens once and gets filed away. It is the input that determines your conformity assessment route, the depth of clinical evidence TGA will expect, and whether overseas certification can be leveraged or whether a TGA-specific pathway is required.

It is also worth noting that Australian classification outcomes can differ from how the same device is classified under EU MDR or other overseas frameworks. A device class assigned under a recognised overseas system does not automatically transfer. The classification work needs to be done against Schedule 2 and Schedule 2A on their own terms, even where the underlying technical file is shared across markets.

Get the classification right early, and the conformity assessment, evidence requirements, and ARTG timeline that follow become predictable. Get it wrong, and those same workstreams become a source of delay later in the process. If you're working through this for your own device, that's exactly where Arcadia can help.